Blame view
mm/hugetlb.c
140 KB
457c89965 treewide: Add SPD... |
1 |
// SPDX-License-Identifier: GPL-2.0-only |
1da177e4c Linux-2.6.12-rc2 |
2 3 |
/* * Generic hugetlb support. |
6d49e352a propagate name ch... |
4 |
* (C) Nadia Yvette Chambers, April 2004 |
1da177e4c Linux-2.6.12-rc2 |
5 |
*/ |
1da177e4c Linux-2.6.12-rc2 |
6 7 |
#include <linux/list.h> #include <linux/init.h> |
1da177e4c Linux-2.6.12-rc2 |
8 |
#include <linux/mm.h> |
e1759c215 proc: switch /pro... |
9 |
#include <linux/seq_file.h> |
1da177e4c Linux-2.6.12-rc2 |
10 11 |
#include <linux/sysctl.h> #include <linux/highmem.h> |
cddb8a5c1 mmu-notifiers: core |
12 |
#include <linux/mmu_notifier.h> |
1da177e4c Linux-2.6.12-rc2 |
13 |
#include <linux/nodemask.h> |
63551ae0f [PATCH] Hugepage ... |
14 |
#include <linux/pagemap.h> |
5da7ca860 [PATCH] Add NUMA ... |
15 |
#include <linux/mempolicy.h> |
3b32123d7 mm: use macros fr... |
16 |
#include <linux/compiler.h> |
aea47ff36 [PATCH] mm: make ... |
17 |
#include <linux/cpuset.h> |
3935baa9b [PATCH] hugepage:... |
18 |
#include <linux/mutex.h> |
97ad1087e memblock: replace... |
19 |
#include <linux/memblock.h> |
a34378701 hugetlb: new sysf... |
20 |
#include <linux/sysfs.h> |
5a0e3ad6a include cleanup: ... |
21 |
#include <linux/slab.h> |
63489f8e8 hugetlbfs: check ... |
22 |
#include <linux/mmdebug.h> |
174cd4b1e sched/headers: Pr... |
23 |
#include <linux/sched/signal.h> |
0fe6e20b9 hugetlb, rmap: ad... |
24 |
#include <linux/rmap.h> |
c6247f72d mm/hugetlb.c: rep... |
25 |
#include <linux/string_helpers.h> |
fd6a03edd HWPOISON, hugetlb... |
26 27 |
#include <linux/swap.h> #include <linux/swapops.h> |
8382d914e mm, hugetlb: impr... |
28 |
#include <linux/jhash.h> |
98fa15f34 mm: replace all o... |
29 |
#include <linux/numa.h> |
3a43ea274 mm/hugetlb: defer... |
30 |
#include <linux/llist.h> |
d6606683a Revert duplicate ... |
31 |
|
63551ae0f [PATCH] Hugepage ... |
32 33 |
#include <asm/page.h> #include <asm/pgtable.h> |
24669e584 hugetlb: use mmu_... |
34 |
#include <asm/tlb.h> |
63551ae0f [PATCH] Hugepage ... |
35 |
|
24669e584 hugetlb: use mmu_... |
36 |
#include <linux/io.h> |
63551ae0f [PATCH] Hugepage ... |
37 |
#include <linux/hugetlb.h> |
9dd540e23 hugetlb/cgroup: a... |
38 |
#include <linux/hugetlb_cgroup.h> |
9a3052306 hugetlb: add per ... |
39 |
#include <linux/node.h> |
1a1aad8a9 userfaultfd: huge... |
40 |
#include <linux/userfaultfd_k.h> |
ab5ac90ae mm, hugetlb: do n... |
41 |
#include <linux/page_owner.h> |
7835e98b2 [PATCH] remove se... |
42 |
#include "internal.h" |
1da177e4c Linux-2.6.12-rc2 |
43 |
|
c3f38a387 hugetlb: make som... |
44 |
int hugetlb_max_hstate __read_mostly; |
e5ff21594 hugetlb: multiple... |
45 46 |
unsigned int default_hstate_idx; struct hstate hstates[HUGE_MAX_HSTATE]; |
641844f56 mm/hugetlb: intro... |
47 48 49 50 51 |
/* * Minimum page order among possible hugepage sizes, set to a proper value * at boot time. */ static unsigned int minimum_order __read_mostly = UINT_MAX; |
e5ff21594 hugetlb: multiple... |
52 |
|
53ba51d21 hugetlb: allow ar... |
53 |
__initdata LIST_HEAD(huge_boot_pages); |
e5ff21594 hugetlb: multiple... |
54 55 56 |
/* for command line parsing */ static struct hstate * __initdata parsed_hstate; static unsigned long __initdata default_hstate_max_huge_pages; |
e11bfbfcb hugetlb: override... |
57 |
static unsigned long __initdata default_hstate_size; |
9fee021d1 mm/hugetlb: intro... |
58 |
static bool __initdata parsed_valid_hugepagesz = true; |
e5ff21594 hugetlb: multiple... |
59 |
|
3935baa9b [PATCH] hugepage:... |
60 |
/* |
31caf665e mm: migrate: make... |
61 62 |
* Protects updates to hugepage_freelists, hugepage_activelist, nr_huge_pages, * free_huge_pages, and surplus_huge_pages. |
3935baa9b [PATCH] hugepage:... |
63 |
*/ |
c3f38a387 hugetlb: make som... |
64 |
DEFINE_SPINLOCK(hugetlb_lock); |
0bd0f9fb1 [PATCH] hugetlb: ... |
65 |
|
8382d914e mm, hugetlb: impr... |
66 67 68 69 70 |
/* * 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; |
c672c7f29 mm/hugetlb: expos... |
71 |
struct mutex *hugetlb_fault_mutex_table ____cacheline_aligned_in_smp; |
8382d914e mm, hugetlb: impr... |
72 |
|
7ca02d0ae hugetlbfs: accept... |
73 74 |
/* Forward declaration */ static int hugetlb_acct_memory(struct hstate *h, long delta); |
90481622d hugepages: fix us... |
75 76 77 78 79 80 81 |
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 |
7ca02d0ae hugetlbfs: accept... |
82 83 84 85 86 87 |
* remain, give up any reservations mased on minimum size and * free the subpool */ if (free) { if (spool->min_hpages != -1) hugetlb_acct_memory(spool->hstate, -spool->min_hpages); |
90481622d hugepages: fix us... |
88 |
kfree(spool); |
7ca02d0ae hugetlbfs: accept... |
89 |
} |
90481622d hugepages: fix us... |
90 |
} |
7ca02d0ae hugetlbfs: accept... |
91 92 |
struct hugepage_subpool *hugepage_new_subpool(struct hstate *h, long max_hpages, long min_hpages) |
90481622d hugepages: fix us... |
93 94 |
{ struct hugepage_subpool *spool; |
c6a918200 hugetlbfs: add mi... |
95 |
spool = kzalloc(sizeof(*spool), GFP_KERNEL); |
90481622d hugepages: fix us... |
96 97 98 99 100 |
if (!spool) return NULL; spin_lock_init(&spool->lock); spool->count = 1; |
7ca02d0ae hugetlbfs: accept... |
101 102 103 104 105 106 107 108 109 |
spool->max_hpages = max_hpages; spool->hstate = h; spool->min_hpages = min_hpages; if (min_hpages != -1 && hugetlb_acct_memory(h, min_hpages)) { kfree(spool); return NULL; } spool->rsv_hpages = min_hpages; |
90481622d hugepages: fix us... |
110 111 112 113 114 115 116 117 118 119 120 |
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); } |
1c5ecae3a hugetlbfs: add mi... |
121 122 123 124 125 126 127 128 129 |
/* * Subpool accounting for allocating and reserving pages. * Return -ENOMEM if there are not enough resources to satisfy the * the request. Otherwise, return the number of pages by which the * global pools must be adjusted (upward). The returned value may * only be different than the passed value (delta) in the case where * a subpool minimum size must be manitained. */ static long hugepage_subpool_get_pages(struct hugepage_subpool *spool, |
90481622d hugepages: fix us... |
130 131 |
long delta) { |
1c5ecae3a hugetlbfs: add mi... |
132 |
long ret = delta; |
90481622d hugepages: fix us... |
133 134 |
if (!spool) |
1c5ecae3a hugetlbfs: add mi... |
135 |
return ret; |
90481622d hugepages: fix us... |
136 137 |
spin_lock(&spool->lock); |
1c5ecae3a hugetlbfs: add mi... |
138 139 140 141 142 143 144 145 |
if (spool->max_hpages != -1) { /* maximum size accounting */ if ((spool->used_hpages + delta) <= spool->max_hpages) spool->used_hpages += delta; else { ret = -ENOMEM; goto unlock_ret; } |
90481622d hugepages: fix us... |
146 |
} |
90481622d hugepages: fix us... |
147 |
|
09a95e29c mm/hugetlb: optim... |
148 149 |
/* minimum size accounting */ if (spool->min_hpages != -1 && spool->rsv_hpages) { |
1c5ecae3a hugetlbfs: add mi... |
150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 |
if (delta > spool->rsv_hpages) { /* * Asking for more reserves than those already taken on * behalf of subpool. Return difference. */ ret = delta - spool->rsv_hpages; spool->rsv_hpages = 0; } else { ret = 0; /* reserves already accounted for */ spool->rsv_hpages -= delta; } } unlock_ret: spin_unlock(&spool->lock); |
90481622d hugepages: fix us... |
165 166 |
return ret; } |
1c5ecae3a hugetlbfs: add mi... |
167 168 169 170 171 172 173 |
/* * Subpool accounting for freeing and unreserving pages. * Return the number of global page reservations that must be dropped. * The return value may only be different than the passed value (delta) * in the case where a subpool minimum size must be maintained. */ static long hugepage_subpool_put_pages(struct hugepage_subpool *spool, |
90481622d hugepages: fix us... |
174 175 |
long delta) { |
1c5ecae3a hugetlbfs: add mi... |
176 |
long ret = delta; |
90481622d hugepages: fix us... |
177 |
if (!spool) |
1c5ecae3a hugetlbfs: add mi... |
178 |
return delta; |
90481622d hugepages: fix us... |
179 180 |
spin_lock(&spool->lock); |
1c5ecae3a hugetlbfs: add mi... |
181 182 183 |
if (spool->max_hpages != -1) /* maximum size accounting */ spool->used_hpages -= delta; |
09a95e29c mm/hugetlb: optim... |
184 185 |
/* minimum size accounting */ if (spool->min_hpages != -1 && spool->used_hpages < spool->min_hpages) { |
1c5ecae3a hugetlbfs: add mi... |
186 187 188 189 190 191 192 193 194 195 196 197 198 199 |
if (spool->rsv_hpages + delta <= spool->min_hpages) ret = 0; else ret = spool->rsv_hpages + delta - spool->min_hpages; spool->rsv_hpages += delta; if (spool->rsv_hpages > spool->min_hpages) spool->rsv_hpages = spool->min_hpages; } /* * If hugetlbfs_put_super couldn't free spool due to an outstanding * quota reference, free it now. */ |
90481622d hugepages: fix us... |
200 |
unlock_or_release_subpool(spool); |
1c5ecae3a hugetlbfs: add mi... |
201 202 |
return ret; |
90481622d hugepages: fix us... |
203 204 205 206 207 208 209 210 211 |
} 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 new helper: file_... |
212 |
return subpool_inode(file_inode(vma->vm_file)); |
90481622d hugepages: fix us... |
213 |
} |
e7c4b0bfd huge page private... |
214 |
/* |
968229048 hugetlb: move res... |
215 216 |
* Region tracking -- allows tracking of reservations and instantiated pages * across the pages in a mapping. |
84afd99b8 hugetlb reservati... |
217 |
* |
1dd308a7b mm/hugetlb: docum... |
218 219 220 221 222 223 224 225 226 227 228 229 230 231 |
* The region data structures are embedded into a resv_map and protected * by a resv_map's lock. The set of regions within the resv_map represent * reservations for huge pages, or huge pages that have already been * instantiated within the map. The from and to elements are huge page * indicies into the associated mapping. from indicates the starting index * of the region. to represents the first index past the end of the region. * * For example, a file region structure with from == 0 and to == 4 represents * four huge pages in a mapping. It is important to note that the to element * represents the first element past the end of the region. This is used in * arithmetic as 4(to) - 0(from) = 4 huge pages in the region. * * Interval notation of the form [from, to) will be used to indicate that * the endpoint from is inclusive and to is exclusive. |
968229048 hugetlb: move res... |
232 233 234 235 236 237 |
*/ struct file_region { struct list_head link; long from; long to; }; |
1dd308a7b mm/hugetlb: docum... |
238 239 |
/* * Add the huge page range represented by [f, t) to the reserve |
5e9113731 mm/hugetlb: add c... |
240 241 242 243 244 245 246 247 |
* map. In the normal case, existing regions will be expanded * to accommodate the specified range. Sufficient regions should * exist for expansion due to the previous call to region_chg * with the same range. However, it is possible that region_del * could have been called after region_chg and modifed the map * in such a way that no region exists to be expanded. In this * case, pull a region descriptor from the cache associated with * the map and use that for the new range. |
cf3ad20bf mm/hugetlb: compu... |
248 249 250 |
* * Return the number of new huge pages added to the map. This * number is greater than or equal to zero. |
1dd308a7b mm/hugetlb: docum... |
251 |
*/ |
1406ec9ba mm, hugetlb: impr... |
252 |
static long region_add(struct resv_map *resv, long f, long t) |
968229048 hugetlb: move res... |
253 |
{ |
1406ec9ba mm, hugetlb: impr... |
254 |
struct list_head *head = &resv->regions; |
968229048 hugetlb: move res... |
255 |
struct file_region *rg, *nrg, *trg; |
cf3ad20bf mm/hugetlb: compu... |
256 |
long add = 0; |
968229048 hugetlb: move res... |
257 |
|
7b24d8616 mm, hugetlb: fix ... |
258 |
spin_lock(&resv->lock); |
968229048 hugetlb: move res... |
259 260 261 262 |
/* Locate the region we are either in or before. */ list_for_each_entry(rg, head, link) if (f <= rg->to) break; |
5e9113731 mm/hugetlb: add c... |
263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 |
/* * If no region exists which can be expanded to include the * specified range, the list must have been modified by an * interleving call to region_del(). Pull a region descriptor * from the cache and use it for this range. */ if (&rg->link == head || t < rg->from) { VM_BUG_ON(resv->region_cache_count <= 0); resv->region_cache_count--; nrg = list_first_entry(&resv->region_cache, struct file_region, link); list_del(&nrg->link); nrg->from = f; nrg->to = t; list_add(&nrg->link, rg->link.prev); add += t - f; goto out_locked; } |
968229048 hugetlb: move res... |
284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 |
/* 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) { |
cf3ad20bf mm/hugetlb: compu... |
302 303 304 305 306 |
/* Decrement return value by the deleted range. * Another range will span this area so that by * end of routine add will be >= zero */ add -= (rg->to - rg->from); |
968229048 hugetlb: move res... |
307 308 309 310 |
list_del(&rg->link); kfree(rg); } } |
cf3ad20bf mm/hugetlb: compu... |
311 312 |
add += (nrg->from - f); /* Added to beginning of region */ |
968229048 hugetlb: move res... |
313 |
nrg->from = f; |
cf3ad20bf mm/hugetlb: compu... |
314 |
add += t - nrg->to; /* Added to end of region */ |
968229048 hugetlb: move res... |
315 |
nrg->to = t; |
cf3ad20bf mm/hugetlb: compu... |
316 |
|
5e9113731 mm/hugetlb: add c... |
317 318 |
out_locked: resv->adds_in_progress--; |
7b24d8616 mm, hugetlb: fix ... |
319 |
spin_unlock(&resv->lock); |
cf3ad20bf mm/hugetlb: compu... |
320 321 |
VM_BUG_ON(add < 0); return add; |
968229048 hugetlb: move res... |
322 |
} |
1dd308a7b mm/hugetlb: docum... |
323 324 325 326 327 328 329 330 331 332 333 334 335 |
/* * Examine the existing reserve map and determine how many * huge pages in the specified range [f, t) are NOT currently * represented. This routine is called before a subsequent * call to region_add that will actually modify the reserve * map to add the specified range [f, t). region_chg does * not change the number of huge pages represented by the * map. However, if the existing regions in the map can not * be expanded to represent the new range, a new file_region * structure is added to the map as a placeholder. This is * so that the subsequent region_add call will have all the * regions it needs and will not fail. * |
5e9113731 mm/hugetlb: add c... |
336 337 338 339 340 341 342 343 |
* Upon entry, region_chg will also examine the cache of region descriptors * associated with the map. If there are not enough descriptors cached, one * will be allocated for the in progress add operation. * * Returns the number of huge pages that need to be added to the existing * reservation map for the range [f, t). This number is greater or equal to * zero. -ENOMEM is returned if a new file_region structure or cache entry * is needed and can not be allocated. |
1dd308a7b mm/hugetlb: docum... |
344 |
*/ |
1406ec9ba mm, hugetlb: impr... |
345 |
static long region_chg(struct resv_map *resv, long f, long t) |
968229048 hugetlb: move res... |
346 |
{ |
1406ec9ba mm, hugetlb: impr... |
347 |
struct list_head *head = &resv->regions; |
7b24d8616 mm, hugetlb: fix ... |
348 |
struct file_region *rg, *nrg = NULL; |
968229048 hugetlb: move res... |
349 |
long chg = 0; |
7b24d8616 mm, hugetlb: fix ... |
350 351 |
retry: spin_lock(&resv->lock); |
5e9113731 mm/hugetlb: add c... |
352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 |
retry_locked: resv->adds_in_progress++; /* * Check for sufficient descriptors in the cache to accommodate * the number of in progress add operations. */ if (resv->adds_in_progress > resv->region_cache_count) { struct file_region *trg; VM_BUG_ON(resv->adds_in_progress - resv->region_cache_count > 1); /* Must drop lock to allocate a new descriptor. */ resv->adds_in_progress--; spin_unlock(&resv->lock); trg = kmalloc(sizeof(*trg), GFP_KERNEL); |
dbe409e4f mm/hugetlb.c: fix... |
368 369 |
if (!trg) { kfree(nrg); |
5e9113731 mm/hugetlb: add c... |
370 |
return -ENOMEM; |
dbe409e4f mm/hugetlb.c: fix... |
371 |
} |
5e9113731 mm/hugetlb: add c... |
372 373 374 375 376 377 |
spin_lock(&resv->lock); list_add(&trg->link, &resv->region_cache); resv->region_cache_count++; goto retry_locked; } |
968229048 hugetlb: move res... |
378 379 380 381 382 383 384 385 386 |
/* 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 mm, hugetlb: fix ... |
387 |
if (!nrg) { |
5e9113731 mm/hugetlb: add c... |
388 |
resv->adds_in_progress--; |
7b24d8616 mm, hugetlb: fix ... |
389 390 391 392 393 394 395 396 397 398 |
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 hugetlb: move res... |
399 |
|
7b24d8616 mm, hugetlb: fix ... |
400 401 402 |
list_add(&nrg->link, rg->link.prev); chg = t - f; goto out_nrg; |
968229048 hugetlb: move res... |
403 404 405 406 407 408 409 410 411 412 413 414 |
} /* 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 mm, hugetlb: fix ... |
415 |
goto out; |
968229048 hugetlb: move res... |
416 |
|
25985edce Fix common misspe... |
417 |
/* We overlap with this area, if it extends further than |
968229048 hugetlb: move res... |
418 419 420 421 422 423 424 425 |
* 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 mm, hugetlb: fix ... |
426 427 428 429 430 431 432 433 |
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 hugetlb: move res... |
434 435 |
return chg; } |
1dd308a7b mm/hugetlb: docum... |
436 |
/* |
5e9113731 mm/hugetlb: add c... |
437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 |
* Abort the in progress add operation. The adds_in_progress field * of the resv_map keeps track of the operations in progress between * calls to region_chg and region_add. Operations are sometimes * aborted after the call to region_chg. In such cases, region_abort * is called to decrement the adds_in_progress counter. * * NOTE: The range arguments [f, t) are not needed or used in this * routine. They are kept to make reading the calling code easier as * arguments will match the associated region_chg call. */ static void region_abort(struct resv_map *resv, long f, long t) { spin_lock(&resv->lock); VM_BUG_ON(!resv->region_cache_count); resv->adds_in_progress--; spin_unlock(&resv->lock); } /* |
feba16e25 mm/hugetlb: add r... |
456 457 458 459 460 461 462 463 464 465 466 467 |
* Delete the specified range [f, t) from the reserve map. If the * t parameter is LONG_MAX, this indicates that ALL regions after f * should be deleted. Locate the regions which intersect [f, t) * and either trim, delete or split the existing regions. * * Returns the number of huge pages deleted from the reserve map. * In the normal case, the return value is zero or more. In the * case where a region must be split, a new region descriptor must * be allocated. If the allocation fails, -ENOMEM will be returned. * NOTE: If the parameter t == LONG_MAX, then we will never split * a region and possibly return -ENOMEM. Callers specifying * t == LONG_MAX do not need to check for -ENOMEM error. |
1dd308a7b mm/hugetlb: docum... |
468 |
*/ |
feba16e25 mm/hugetlb: add r... |
469 |
static long region_del(struct resv_map *resv, long f, long t) |
968229048 hugetlb: move res... |
470 |
{ |
1406ec9ba mm, hugetlb: impr... |
471 |
struct list_head *head = &resv->regions; |
968229048 hugetlb: move res... |
472 |
struct file_region *rg, *trg; |
feba16e25 mm/hugetlb: add r... |
473 474 |
struct file_region *nrg = NULL; long del = 0; |
968229048 hugetlb: move res... |
475 |
|
feba16e25 mm/hugetlb: add r... |
476 |
retry: |
7b24d8616 mm, hugetlb: fix ... |
477 |
spin_lock(&resv->lock); |
feba16e25 mm/hugetlb: add r... |
478 |
list_for_each_entry_safe(rg, trg, head, link) { |
dbe409e4f mm/hugetlb.c: fix... |
479 480 481 482 483 484 485 486 |
/* * Skip regions before the range to be deleted. file_region * ranges are normally of the form [from, to). However, there * may be a "placeholder" entry in the map which is of the form * (from, to) with from == to. Check for placeholder entries * at the beginning of the range to be deleted. */ if (rg->to <= f && (rg->to != rg->from || rg->to != f)) |
feba16e25 mm/hugetlb: add r... |
487 |
continue; |
dbe409e4f mm/hugetlb.c: fix... |
488 |
|
feba16e25 mm/hugetlb: add r... |
489 |
if (rg->from >= t) |
968229048 hugetlb: move res... |
490 |
break; |
968229048 hugetlb: move res... |
491 |
|
feba16e25 mm/hugetlb: add r... |
492 493 494 495 496 497 498 499 500 501 502 503 504 |
if (f > rg->from && t < rg->to) { /* Must split region */ /* * Check for an entry in the cache before dropping * lock and attempting allocation. */ if (!nrg && resv->region_cache_count > resv->adds_in_progress) { nrg = list_first_entry(&resv->region_cache, struct file_region, link); list_del(&nrg->link); resv->region_cache_count--; } |
968229048 hugetlb: move res... |
505 |
|
feba16e25 mm/hugetlb: add r... |
506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 |
if (!nrg) { spin_unlock(&resv->lock); nrg = kmalloc(sizeof(*nrg), GFP_KERNEL); if (!nrg) return -ENOMEM; goto retry; } del += t - f; /* New entry for end of split region */ nrg->from = t; nrg->to = rg->to; INIT_LIST_HEAD(&nrg->link); /* Original entry is trimmed */ rg->to = f; list_add(&nrg->link, &rg->link); nrg = NULL; |
968229048 hugetlb: move res... |
526 |
break; |
feba16e25 mm/hugetlb: add r... |
527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 |
} if (f <= rg->from && t >= rg->to) { /* Remove entire region */ del += rg->to - rg->from; list_del(&rg->link); kfree(rg); continue; } if (f <= rg->from) { /* Trim beginning of region */ del += t - rg->from; rg->from = t; } else { /* Trim end of region */ del += rg->to - f; rg->to = f; } |
968229048 hugetlb: move res... |
543 |
} |
7b24d8616 mm, hugetlb: fix ... |
544 |
|
7b24d8616 mm, hugetlb: fix ... |
545 |
spin_unlock(&resv->lock); |
feba16e25 mm/hugetlb: add r... |
546 547 |
kfree(nrg); return del; |
968229048 hugetlb: move res... |
548 |
} |
1dd308a7b mm/hugetlb: docum... |
549 |
/* |
b5cec28d3 hugetlbfs: trunca... |
550 551 552 553 554 555 556 557 |
* A rare out of memory error was encountered which prevented removal of * the reserve map region for a page. The huge page itself was free'ed * and removed from the page cache. This routine will adjust the subpool * usage count, and the global reserve count if needed. By incrementing * these counts, the reserve map entry which could not be deleted will * appear as a "reserved" entry instead of simply dangling with incorrect * counts. */ |
72e2936c0 mm: remove unnece... |
558 |
void hugetlb_fix_reserve_counts(struct inode *inode) |
b5cec28d3 hugetlbfs: trunca... |
559 560 561 562 563 |
{ struct hugepage_subpool *spool = subpool_inode(inode); long rsv_adjust; rsv_adjust = hugepage_subpool_get_pages(spool, 1); |
72e2936c0 mm: remove unnece... |
564 |
if (rsv_adjust) { |
b5cec28d3 hugetlbfs: trunca... |
565 566 567 568 569 570 571 |
struct hstate *h = hstate_inode(inode); hugetlb_acct_memory(h, 1); } } /* |
1dd308a7b mm/hugetlb: docum... |
572 573 574 |
* Count and return the number of huge pages in the reserve map * that intersect with the range [f, t). */ |
1406ec9ba mm, hugetlb: impr... |
575 |
static long region_count(struct resv_map *resv, long f, long t) |
84afd99b8 hugetlb reservati... |
576 |
{ |
1406ec9ba mm, hugetlb: impr... |
577 |
struct list_head *head = &resv->regions; |
84afd99b8 hugetlb reservati... |
578 579 |
struct file_region *rg; long chg = 0; |
7b24d8616 mm, hugetlb: fix ... |
580 |
spin_lock(&resv->lock); |
84afd99b8 hugetlb reservati... |
581 582 |
/* Locate each segment we overlap with, and count that overlap. */ list_for_each_entry(rg, head, link) { |
f2135a4a5 mm/hugetlb.c: use... |
583 584 |
long seg_from; long seg_to; |
84afd99b8 hugetlb reservati... |
585 586 587 588 589 590 591 592 593 594 595 |
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 mm, hugetlb: fix ... |
596 |
spin_unlock(&resv->lock); |
84afd99b8 hugetlb reservati... |
597 598 599 |
return chg; } |
968229048 hugetlb: move res... |
600 |
/* |
e7c4b0bfd huge page private... |
601 |
* Convert the address within this vma to the page offset within |
e7c4b0bfd huge page private... |
602 603 |
* the mapping, in pagecache page units; huge pages here. */ |
a55164389 hugetlb: modular ... |
604 605 |
static pgoff_t vma_hugecache_offset(struct hstate *h, struct vm_area_struct *vma, unsigned long address) |
e7c4b0bfd huge page private... |
606 |
{ |
a55164389 hugetlb: modular ... |
607 608 |
return ((address - vma->vm_start) >> huge_page_shift(h)) + (vma->vm_pgoff >> huge_page_order(h)); |
e7c4b0bfd huge page private... |
609 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
610 611 612 613 614 |
pgoff_t linear_hugepage_index(struct vm_area_struct *vma, unsigned long address) { return vma_hugecache_offset(hstate_vma(vma), vma, address); } |
dee410792 /dev/dax, core: f... |
615 |
EXPORT_SYMBOL_GPL(linear_hugepage_index); |
0fe6e20b9 hugetlb, rmap: ad... |
616 |
|
84afd99b8 hugetlb reservati... |
617 |
/* |
08fba6998 mm: report the pa... |
618 619 620 621 622 |
* 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) { |
05ea88608 mm, hugetlbfs: in... |
623 624 625 |
if (vma->vm_ops && vma->vm_ops->pagesize) return vma->vm_ops->pagesize(vma); return PAGE_SIZE; |
08fba6998 mm: report the pa... |
626 |
} |
f340ca0f0 hugetlbfs: export... |
627 |
EXPORT_SYMBOL_GPL(vma_kernel_pagesize); |
08fba6998 mm: report the pa... |
628 629 |
/* |
3340289dd mm: report the MM... |
630 631 |
* 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 |
09135cc59 mm, powerpc: use ... |
632 633 |
* architectures where it differs, an architecture-specific 'strong' * version of this symbol is required. |
3340289dd mm: report the MM... |
634 |
*/ |
09135cc59 mm, powerpc: use ... |
635 |
__weak unsigned long vma_mmu_pagesize(struct vm_area_struct *vma) |
3340289dd mm: report the MM... |
636 637 638 |
{ return vma_kernel_pagesize(vma); } |
3340289dd mm: report the MM... |
639 640 |
/* |
84afd99b8 hugetlb reservati... |
641 642 643 644 645 646 |
* 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 hugetlb: guarante... |
647 |
#define HPAGE_RESV_MASK (HPAGE_RESV_OWNER | HPAGE_RESV_UNMAPPED) |
84afd99b8 hugetlb reservati... |
648 |
|
a1e78772d hugetlb: reserve ... |
649 650 651 652 653 654 655 656 657 |
/* * 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 hugetlb reservati... |
658 659 660 661 662 663 664 665 666 |
* * 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 hugetlb: reserve ... |
667 |
*/ |
e7c4b0bfd huge page private... |
668 669 670 671 672 673 674 675 676 677 |
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 mm, hugetlb: unif... |
678 |
struct resv_map *resv_map_alloc(void) |
84afd99b8 hugetlb reservati... |
679 680 |
{ struct resv_map *resv_map = kmalloc(sizeof(*resv_map), GFP_KERNEL); |
5e9113731 mm/hugetlb: add c... |
681 682 683 684 685 |
struct file_region *rg = kmalloc(sizeof(*rg), GFP_KERNEL); if (!resv_map || !rg) { kfree(resv_map); kfree(rg); |
84afd99b8 hugetlb reservati... |
686 |
return NULL; |
5e9113731 mm/hugetlb: add c... |
687 |
} |
84afd99b8 hugetlb reservati... |
688 689 |
kref_init(&resv_map->refs); |
7b24d8616 mm, hugetlb: fix ... |
690 |
spin_lock_init(&resv_map->lock); |
84afd99b8 hugetlb reservati... |
691 |
INIT_LIST_HEAD(&resv_map->regions); |
5e9113731 mm/hugetlb: add c... |
692 693 694 695 696 |
resv_map->adds_in_progress = 0; INIT_LIST_HEAD(&resv_map->region_cache); list_add(&rg->link, &resv_map->region_cache); resv_map->region_cache_count = 1; |
84afd99b8 hugetlb reservati... |
697 698 |
return resv_map; } |
9119a41e9 mm, hugetlb: unif... |
699 |
void resv_map_release(struct kref *ref) |
84afd99b8 hugetlb reservati... |
700 701 |
{ struct resv_map *resv_map = container_of(ref, struct resv_map, refs); |
5e9113731 mm/hugetlb: add c... |
702 703 |
struct list_head *head = &resv_map->region_cache; struct file_region *rg, *trg; |
84afd99b8 hugetlb reservati... |
704 705 |
/* Clear out any active regions before we release the map. */ |
feba16e25 mm/hugetlb: add r... |
706 |
region_del(resv_map, 0, LONG_MAX); |
5e9113731 mm/hugetlb: add c... |
707 708 709 710 711 712 713 714 |
/* ... and any entries left in the cache */ list_for_each_entry_safe(rg, trg, head, link) { list_del(&rg->link); kfree(rg); } VM_BUG_ON(resv_map->adds_in_progress); |
84afd99b8 hugetlb reservati... |
715 716 |
kfree(resv_map); } |
4e35f4838 mm, hugetlb: use ... |
717 718 |
static inline struct resv_map *inode_resv_map(struct inode *inode) { |
f27a5136f hugetlbfs: always... |
719 720 721 722 723 724 725 726 727 |
/* * At inode evict time, i_mapping may not point to the original * address space within the inode. This original address space * contains the pointer to the resv_map. So, always use the * address space embedded within the inode. * The VERY common case is inode->mapping == &inode->i_data but, * this may not be true for device special inodes. */ return (struct resv_map *)(&inode->i_data)->private_data; |
4e35f4838 mm, hugetlb: use ... |
728 |
} |
84afd99b8 hugetlb reservati... |
729 |
static struct resv_map *vma_resv_map(struct vm_area_struct *vma) |
a1e78772d hugetlb: reserve ... |
730 |
{ |
81d1b09c6 mm: convert a few... |
731 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); |
4e35f4838 mm, hugetlb: use ... |
732 733 734 735 736 737 738 |
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 hugetlb reservati... |
739 740 |
return (struct resv_map *)(get_vma_private_data(vma) & ~HPAGE_RESV_MASK); |
4e35f4838 mm, hugetlb: use ... |
741 |
} |
a1e78772d hugetlb: reserve ... |
742 |
} |
84afd99b8 hugetlb reservati... |
743 |
static void set_vma_resv_map(struct vm_area_struct *vma, struct resv_map *map) |
a1e78772d hugetlb: reserve ... |
744 |
{ |
81d1b09c6 mm: convert a few... |
745 746 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); VM_BUG_ON_VMA(vma->vm_flags & VM_MAYSHARE, vma); |
a1e78772d hugetlb: reserve ... |
747 |
|
84afd99b8 hugetlb reservati... |
748 749 |
set_vma_private_data(vma, (get_vma_private_data(vma) & HPAGE_RESV_MASK) | (unsigned long)map); |
04f2cbe35 hugetlb: guarante... |
750 751 752 753 |
} static void set_vma_resv_flags(struct vm_area_struct *vma, unsigned long flags) { |
81d1b09c6 mm: convert a few... |
754 755 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); VM_BUG_ON_VMA(vma->vm_flags & VM_MAYSHARE, vma); |
e7c4b0bfd huge page private... |
756 757 |
set_vma_private_data(vma, get_vma_private_data(vma) | flags); |
04f2cbe35 hugetlb: guarante... |
758 759 760 761 |
} static int is_vma_resv_set(struct vm_area_struct *vma, unsigned long flag) { |
81d1b09c6 mm: convert a few... |
762 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); |
e7c4b0bfd huge page private... |
763 764 |
return (get_vma_private_data(vma) & flag) != 0; |
a1e78772d hugetlb: reserve ... |
765 |
} |
04f2cbe35 hugetlb: guarante... |
766 |
/* Reset counters to 0 and clear all HPAGE_RESV_* flags */ |
a1e78772d hugetlb: reserve ... |
767 768 |
void reset_vma_resv_huge_pages(struct vm_area_struct *vma) { |
81d1b09c6 mm: convert a few... |
769 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); |
f83a275db mm: account for M... |
770 |
if (!(vma->vm_flags & VM_MAYSHARE)) |
a1e78772d hugetlb: reserve ... |
771 772 773 774 |
vma->vm_private_data = (void *)0; } /* Returns true if the VMA has associated reserve pages */ |
559ec2f8f mm/hugetlb.c: mak... |
775 |
static bool vma_has_reserves(struct vm_area_struct *vma, long chg) |
a1e78772d hugetlb: reserve ... |
776 |
{ |
af0ed73e6 mm, hugetlb: decr... |
777 778 779 780 781 782 783 784 785 786 787 |
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) |
559ec2f8f mm/hugetlb.c: mak... |
788 |
return true; |
af0ed73e6 mm, hugetlb: decr... |
789 |
else |
559ec2f8f mm/hugetlb.c: mak... |
790 |
return false; |
af0ed73e6 mm, hugetlb: decr... |
791 |
} |
a63884e92 mm, hugetlb: remo... |
792 793 |
/* Shared mappings always use reserves */ |
1fb1b0e9e mm/hugetlb: vma_h... |
794 795 796 797 798 799 800 801 802 803 804 805 806 |
if (vma->vm_flags & VM_MAYSHARE) { /* * We know VM_NORESERVE is not set. Therefore, there SHOULD * be a region map for all pages. The only situation where * there is no region map is if a hole was punched via * fallocate. In this case, there really are no reverves to * use. This situation is indicated if chg != 0. */ if (chg) return false; else return true; } |
a63884e92 mm, hugetlb: remo... |
807 808 809 810 811 |
/* * Only the process that called mmap() has reserves for * private mappings. */ |
67961f9db mm/hugetlb: fix h... |
812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 |
if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) { /* * Like the shared case above, a hole punch or truncate * could have been performed on the private mapping. * Examine the value of chg to determine if reserves * actually exist or were previously consumed. * Very Subtle - The value of chg comes from a previous * call to vma_needs_reserves(). The reserve map for * private mappings has different (opposite) semantics * than that of shared mappings. vma_needs_reserves() * has already taken this difference in semantics into * account. Therefore, the meaning of chg is the same * as in the shared case above. Code could easily be * combined, but keeping it separate draws attention to * subtle differences. */ if (chg) return false; else return true; } |
a63884e92 mm, hugetlb: remo... |
833 |
|
559ec2f8f mm/hugetlb.c: mak... |
834 |
return false; |
a1e78772d hugetlb: reserve ... |
835 |
} |
a55164389 hugetlb: modular ... |
836 |
static void enqueue_huge_page(struct hstate *h, struct page *page) |
1da177e4c Linux-2.6.12-rc2 |
837 838 |
{ int nid = page_to_nid(page); |
0edaecfab hugetlb: add a li... |
839 |
list_move(&page->lru, &h->hugepage_freelists[nid]); |
a55164389 hugetlb: modular ... |
840 841 |
h->free_huge_pages++; h->free_huge_pages_node[nid]++; |
1da177e4c Linux-2.6.12-rc2 |
842 |
} |
94310cbca mm/madvise: enabl... |
843 |
static struct page *dequeue_huge_page_node_exact(struct hstate *h, int nid) |
bf50bab2b hugetlb: add allo... |
844 845 |
{ struct page *page; |
c8721bbbd mm: memory-hotplu... |
846 |
list_for_each_entry(page, &h->hugepage_freelists[nid], lru) |
243abd5b7 mm: hugetlb: prev... |
847 |
if (!PageHWPoison(page)) |
c8721bbbd mm: memory-hotplu... |
848 849 850 851 852 853 |
break; /* * if 'non-isolated free hugepage' not found on the list, * the allocation fails. */ if (&h->hugepage_freelists[nid] == &page->lru) |
bf50bab2b hugetlb: add allo... |
854 |
return NULL; |
0edaecfab hugetlb: add a li... |
855 |
list_move(&page->lru, &h->hugepage_activelist); |
a9869b837 hugetlb: move ref... |
856 |
set_page_refcounted(page); |
bf50bab2b hugetlb: add allo... |
857 858 859 860 |
h->free_huge_pages--; h->free_huge_pages_node[nid]--; return page; } |
3e59fcb0e hugetlb: add supp... |
861 862 |
static struct page *dequeue_huge_page_nodemask(struct hstate *h, gfp_t gfp_mask, int nid, nodemask_t *nmask) |
94310cbca mm/madvise: enabl... |
863 |
{ |
3e59fcb0e hugetlb: add supp... |
864 865 866 867 |
unsigned int cpuset_mems_cookie; struct zonelist *zonelist; struct zone *zone; struct zoneref *z; |
98fa15f34 mm: replace all o... |
868 |
int node = NUMA_NO_NODE; |
94310cbca mm/madvise: enabl... |
869 |
|
3e59fcb0e hugetlb: add supp... |
870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 |
zonelist = node_zonelist(nid, gfp_mask); retry_cpuset: cpuset_mems_cookie = read_mems_allowed_begin(); for_each_zone_zonelist_nodemask(zone, z, zonelist, gfp_zone(gfp_mask), nmask) { struct page *page; if (!cpuset_zone_allowed(zone, gfp_mask)) continue; /* * no need to ask again on the same node. Pool is node rather than * zone aware */ if (zone_to_nid(zone) == node) continue; node = zone_to_nid(zone); |
94310cbca mm/madvise: enabl... |
886 |
|
94310cbca mm/madvise: enabl... |
887 888 889 890 |
page = dequeue_huge_page_node_exact(h, node); if (page) return page; } |
3e59fcb0e hugetlb: add supp... |
891 892 |
if (unlikely(read_mems_allowed_retry(cpuset_mems_cookie))) goto retry_cpuset; |
94310cbca mm/madvise: enabl... |
893 894 |
return NULL; } |
86cdb465c mm: prepare to re... |
895 896 897 |
/* Movability of hugepages depends on migration support. */ static inline gfp_t htlb_alloc_mask(struct hstate *h) { |
7ed2c31da mm/hugetlb: disti... |
898 |
if (hugepage_movable_supported(h)) |
86cdb465c mm: prepare to re... |
899 900 901 902 |
return GFP_HIGHUSER_MOVABLE; else return GFP_HIGHUSER; } |
a55164389 hugetlb: modular ... |
903 904 |
static struct page *dequeue_huge_page_vma(struct hstate *h, struct vm_area_struct *vma, |
af0ed73e6 mm, hugetlb: decr... |
905 906 |
unsigned long address, int avoid_reserve, long chg) |
1da177e4c Linux-2.6.12-rc2 |
907 |
{ |
3e59fcb0e hugetlb: add supp... |
908 |
struct page *page; |
480eccf9a Fix NUMA Memory P... |
909 |
struct mempolicy *mpol; |
04ec6264f mm, page_alloc: p... |
910 |
gfp_t gfp_mask; |
3e59fcb0e hugetlb: add supp... |
911 |
nodemask_t *nodemask; |
04ec6264f mm, page_alloc: p... |
912 |
int nid; |
1da177e4c Linux-2.6.12-rc2 |
913 |
|
a1e78772d hugetlb: reserve ... |
914 915 916 917 918 |
/* * 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 mm, hugetlb: decr... |
919 |
if (!vma_has_reserves(vma, chg) && |
a55164389 hugetlb: modular ... |
920 |
h->free_huge_pages - h->resv_huge_pages == 0) |
c0ff7453b cpuset,mm: fix no... |
921 |
goto err; |
a1e78772d hugetlb: reserve ... |
922 |
|
04f2cbe35 hugetlb: guarante... |
923 |
/* If reserves cannot be used, ensure enough pages are in the pool */ |
a55164389 hugetlb: modular ... |
924 |
if (avoid_reserve && h->free_huge_pages - h->resv_huge_pages == 0) |
6eab04a87 treewide: remove ... |
925 |
goto err; |
04f2cbe35 hugetlb: guarante... |
926 |
|
04ec6264f mm, page_alloc: p... |
927 928 |
gfp_mask = htlb_alloc_mask(h); nid = huge_node(vma, address, gfp_mask, &mpol, &nodemask); |
3e59fcb0e hugetlb: add supp... |
929 930 931 932 |
page = dequeue_huge_page_nodemask(h, gfp_mask, nid, nodemask); if (page && !avoid_reserve && vma_has_reserves(vma, chg)) { SetPagePrivate(page); h->resv_huge_pages--; |
1da177e4c Linux-2.6.12-rc2 |
933 |
} |
cc9a6c877 cpuset: mm: reduc... |
934 |
|
52cd3b074 mempolicy: rework... |
935 |
mpol_cond_put(mpol); |
1da177e4c Linux-2.6.12-rc2 |
936 |
return page; |
cc9a6c877 cpuset: mm: reduc... |
937 938 |
err: |
cc9a6c877 cpuset: mm: reduc... |
939 |
return NULL; |
1da177e4c Linux-2.6.12-rc2 |
940 |
} |
1cac6f2c0 hugetlb: move hel... |
941 942 943 944 945 946 947 948 949 |
/* * 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) { |
0edaf86cf include/linux/nod... |
950 |
nid = next_node_in(nid, *nodes_allowed); |
1cac6f2c0 hugetlb: move hel... |
951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 |
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--) |
e1073d1e7 mm/hugetlb: clean... |
1011 |
#ifdef CONFIG_ARCH_HAS_GIGANTIC_PAGE |
944d9fec8 hugetlb: add supp... |
1012 |
static void destroy_compound_gigantic_page(struct page *page, |
d00181b96 mm: use 'unsigned... |
1013 |
unsigned int order) |
944d9fec8 hugetlb: add supp... |
1014 1015 1016 1017 |
{ int i; int nr_pages = 1 << order; struct page *p = page + 1; |
c8cc708a3 mm/hugetlb: clear... |
1018 |
atomic_set(compound_mapcount_ptr(page), 0); |
944d9fec8 hugetlb: add supp... |
1019 |
for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) { |
1d798ca3f mm: make compound... |
1020 |
clear_compound_head(p); |
944d9fec8 hugetlb: add supp... |
1021 |
set_page_refcounted(p); |
944d9fec8 hugetlb: add supp... |
1022 1023 1024 1025 1026 |
} set_compound_order(page, 0); __ClearPageHead(page); } |
d00181b96 mm: use 'unsigned... |
1027 |
static void free_gigantic_page(struct page *page, unsigned int order) |
944d9fec8 hugetlb: add supp... |
1028 1029 1030 |
{ free_contig_range(page_to_pfn(page), 1 << order); } |
4eb0716e8 hugetlb: allow to... |
1031 |
#ifdef CONFIG_CONTIG_ALLOC |
944d9fec8 hugetlb: add supp... |
1032 |
static int __alloc_gigantic_page(unsigned long start_pfn, |
79b63f12a mm, hugetlb: do n... |
1033 |
unsigned long nr_pages, gfp_t gfp_mask) |
944d9fec8 hugetlb: add supp... |
1034 1035 |
{ unsigned long end_pfn = start_pfn + nr_pages; |
ca96b6253 mm: alloc_contig_... |
1036 |
return alloc_contig_range(start_pfn, end_pfn, MIGRATE_MOVABLE, |
79b63f12a mm, hugetlb: do n... |
1037 |
gfp_mask); |
944d9fec8 hugetlb: add supp... |
1038 |
} |
f44b2dda8 mm/hugetlb: add s... |
1039 1040 |
static bool pfn_range_valid_gigantic(struct zone *z, unsigned long start_pfn, unsigned long nr_pages) |
944d9fec8 hugetlb: add supp... |
1041 1042 1043 1044 1045 |
{ unsigned long i, end_pfn = start_pfn + nr_pages; struct page *page; for (i = start_pfn; i < end_pfn; i++) { |
f231fe423 hugetlbfs: don't ... |
1046 1047 |
page = pfn_to_online_page(i); if (!page) |
944d9fec8 hugetlb: add supp... |
1048 |
return false; |
f44b2dda8 mm/hugetlb: add s... |
1049 1050 |
if (page_zone(page) != z) return false; |
944d9fec8 hugetlb: add supp... |
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 |
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); } |
d9cc948f6 mm, hugetlb: inte... |
1070 1071 |
static struct page *alloc_gigantic_page(struct hstate *h, gfp_t gfp_mask, int nid, nodemask_t *nodemask) |
944d9fec8 hugetlb: add supp... |
1072 |
{ |
79b63f12a mm, hugetlb: do n... |
1073 |
unsigned int order = huge_page_order(h); |
944d9fec8 hugetlb: add supp... |
1074 1075 |
unsigned long nr_pages = 1 << order; unsigned long ret, pfn, flags; |
79b63f12a mm, hugetlb: do n... |
1076 1077 1078 |
struct zonelist *zonelist; struct zone *zone; struct zoneref *z; |
944d9fec8 hugetlb: add supp... |
1079 |
|
79b63f12a mm, hugetlb: do n... |
1080 |
zonelist = node_zonelist(nid, gfp_mask); |
d9cc948f6 mm, hugetlb: inte... |
1081 |
for_each_zone_zonelist_nodemask(zone, z, zonelist, gfp_zone(gfp_mask), nodemask) { |
79b63f12a mm, hugetlb: do n... |
1082 |
spin_lock_irqsave(&zone->lock, flags); |
944d9fec8 hugetlb: add supp... |
1083 |
|
79b63f12a mm, hugetlb: do n... |
1084 1085 1086 |
pfn = ALIGN(zone->zone_start_pfn, nr_pages); while (zone_spans_last_pfn(zone, pfn, nr_pages)) { if (pfn_range_valid_gigantic(zone, pfn, nr_pages)) { |
944d9fec8 hugetlb: add supp... |
1087 1088 1089 1090 1091 1092 1093 |
/* * 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... */ |
79b63f12a mm, hugetlb: do n... |
1094 1095 |
spin_unlock_irqrestore(&zone->lock, flags); ret = __alloc_gigantic_page(pfn, nr_pages, gfp_mask); |
944d9fec8 hugetlb: add supp... |
1096 1097 |
if (!ret) return pfn_to_page(pfn); |
79b63f12a mm, hugetlb: do n... |
1098 |
spin_lock_irqsave(&zone->lock, flags); |
944d9fec8 hugetlb: add supp... |
1099 1100 1101 |
} pfn += nr_pages; } |
79b63f12a mm, hugetlb: do n... |
1102 |
spin_unlock_irqrestore(&zone->lock, flags); |
944d9fec8 hugetlb: add supp... |
1103 1104 1105 1106 1107 1108 |
} return NULL; } static void prep_new_huge_page(struct hstate *h, struct page *page, int nid); |
d00181b96 mm: use 'unsigned... |
1109 |
static void prep_compound_gigantic_page(struct page *page, unsigned int order); |
4eb0716e8 hugetlb: allow to... |
1110 1111 1112 1113 1114 1115 1116 |
#else /* !CONFIG_CONTIG_ALLOC */ static struct page *alloc_gigantic_page(struct hstate *h, gfp_t gfp_mask, int nid, nodemask_t *nodemask) { return NULL; } #endif /* CONFIG_CONTIG_ALLOC */ |
944d9fec8 hugetlb: add supp... |
1117 |
|
e1073d1e7 mm/hugetlb: clean... |
1118 |
#else /* !CONFIG_ARCH_HAS_GIGANTIC_PAGE */ |
d9cc948f6 mm, hugetlb: inte... |
1119 |
static struct page *alloc_gigantic_page(struct hstate *h, gfp_t gfp_mask, |
4eb0716e8 hugetlb: allow to... |
1120 1121 1122 1123 |
int nid, nodemask_t *nodemask) { return NULL; } |
d00181b96 mm: use 'unsigned... |
1124 |
static inline void free_gigantic_page(struct page *page, unsigned int order) { } |
944d9fec8 hugetlb: add supp... |
1125 |
static inline void destroy_compound_gigantic_page(struct page *page, |
d00181b96 mm: use 'unsigned... |
1126 |
unsigned int order) { } |
944d9fec8 hugetlb: add supp... |
1127 |
#endif |
a55164389 hugetlb: modular ... |
1128 |
static void update_and_free_page(struct hstate *h, struct page *page) |
6af2acb66 hugetlb: Move upd... |
1129 1130 |
{ int i; |
a55164389 hugetlb: modular ... |
1131 |
|
4eb0716e8 hugetlb: allow to... |
1132 |
if (hstate_is_gigantic(h) && !gigantic_page_runtime_supported()) |
944d9fec8 hugetlb: add supp... |
1133 |
return; |
18229df5b hugetlb: pull gig... |
1134 |
|
a55164389 hugetlb: modular ... |
1135 1136 1137 |
h->nr_huge_pages--; h->nr_huge_pages_node[page_to_nid(page)]--; for (i = 0; i < pages_per_huge_page(h); i++) { |
32f84528f mm: hugetlb: fix ... |
1138 1139 |
page[i].flags &= ~(1 << PG_locked | 1 << PG_error | 1 << PG_referenced | 1 << PG_dirty | |
a7407a27c hugetlb: update_a... |
1140 1141 |
1 << PG_active | 1 << PG_private | 1 << PG_writeback); |
6af2acb66 hugetlb: Move upd... |
1142 |
} |
309381fea mm: dump page whe... |
1143 |
VM_BUG_ON_PAGE(hugetlb_cgroup_from_page(page), page); |
f1e61557f mm: pack compound... |
1144 |
set_compound_page_dtor(page, NULL_COMPOUND_DTOR); |
6af2acb66 hugetlb: Move upd... |
1145 |
set_page_refcounted(page); |
944d9fec8 hugetlb: add supp... |
1146 1147 1148 1149 |
if (hstate_is_gigantic(h)) { destroy_compound_gigantic_page(page, huge_page_order(h)); free_gigantic_page(page, huge_page_order(h)); } else { |
944d9fec8 hugetlb: add supp... |
1150 1151 |
__free_pages(page, huge_page_order(h)); } |
6af2acb66 hugetlb: Move upd... |
1152 |
} |
e5ff21594 hugetlb: multiple... |
1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 |
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; } |
bcc542223 mm: hugetlb: intr... |
1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 |
/* * Test to determine whether the hugepage is "active/in-use" (i.e. being linked * to hstate->hugepage_activelist.) * * This function can be called for tail pages, but never returns true for them. */ bool page_huge_active(struct page *page) { VM_BUG_ON_PAGE(!PageHuge(page), page); return PageHead(page) && PagePrivate(&page[1]); } /* never called for tail page */ static void set_page_huge_active(struct page *page) { VM_BUG_ON_PAGE(!PageHeadHuge(page), page); SetPagePrivate(&page[1]); } static void clear_page_huge_active(struct page *page) { VM_BUG_ON_PAGE(!PageHeadHuge(page), page); ClearPagePrivate(&page[1]); } |
ab5ac90ae mm, hugetlb: do n... |
1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 |
/* * Internal hugetlb specific page flag. Do not use outside of the hugetlb * code */ static inline bool PageHugeTemporary(struct page *page) { if (!PageHuge(page)) return false; return (unsigned long)page[2].mapping == -1U; } static inline void SetPageHugeTemporary(struct page *page) { page[2].mapping = (void *)-1U; } static inline void ClearPageHugeTemporary(struct page *page) { page[2].mapping = NULL; } |
3a43ea274 mm/hugetlb: defer... |
1208 |
static void __free_huge_page(struct page *page) |
27a85ef1b [PATCH] hugepage:... |
1209 |
{ |
a55164389 hugetlb: modular ... |
1210 1211 1212 1213 |
/* * Can't pass hstate in here because it is called from the * compound page destructor. */ |
e5ff21594 hugetlb: multiple... |
1214 |
struct hstate *h = page_hstate(page); |
7893d1d50 hugetlb: Try to g... |
1215 |
int nid = page_to_nid(page); |
90481622d hugepages: fix us... |
1216 1217 |
struct hugepage_subpool *spool = (struct hugepage_subpool *)page_private(page); |
07443a85a mm, hugetlb: retu... |
1218 |
bool restore_reserve; |
27a85ef1b [PATCH] hugepage:... |
1219 |
|
b4330afbe mm/hugetlb: fix g... |
1220 1221 |
VM_BUG_ON_PAGE(page_count(page), page); VM_BUG_ON_PAGE(page_mapcount(page), page); |
8ace22bce hugetlbfs: call V... |
1222 1223 1224 |
set_page_private(page, 0); page->mapping = NULL; |
07443a85a mm, hugetlb: retu... |
1225 |
restore_reserve = PagePrivate(page); |
16c794b4f mm/hugetlb.c: cor... |
1226 |
ClearPagePrivate(page); |
27a85ef1b [PATCH] hugepage:... |
1227 |
|
1c5ecae3a hugetlbfs: add mi... |
1228 |
/* |
0919e1b69 hugetlbfs: on res... |
1229 1230 1231 1232 1233 1234 |
* If PagePrivate() was set on page, page allocation consumed a * reservation. If the page was associated with a subpool, there * would have been a page reserved in the subpool before allocation * via hugepage_subpool_get_pages(). Since we are 'restoring' the * reservtion, do not call hugepage_subpool_put_pages() as this will * remove the reserved page from the subpool. |
1c5ecae3a hugetlbfs: add mi... |
1235 |
*/ |
0919e1b69 hugetlbfs: on res... |
1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 |
if (!restore_reserve) { /* * A return code of zero implies that the subpool will be * under its minimum size if the reservation is not restored * after page is free. Therefore, force restore_reserve * operation. */ if (hugepage_subpool_put_pages(spool, 1) == 0) restore_reserve = true; } |
1c5ecae3a hugetlbfs: add mi... |
1246 |
|
27a85ef1b [PATCH] hugepage:... |
1247 |
spin_lock(&hugetlb_lock); |
bcc542223 mm: hugetlb: intr... |
1248 |
clear_page_huge_active(page); |
6d76dcf40 hugetlb/cgroup: a... |
1249 1250 |
hugetlb_cgroup_uncharge_page(hstate_index(h), pages_per_huge_page(h), page); |
07443a85a mm, hugetlb: retu... |
1251 1252 |
if (restore_reserve) h->resv_huge_pages++; |
ab5ac90ae mm, hugetlb: do n... |
1253 1254 1255 1256 1257 |
if (PageHugeTemporary(page)) { list_del(&page->lru); ClearPageHugeTemporary(page); update_and_free_page(h, page); } else if (h->surplus_huge_pages_node[nid]) { |
0edaecfab hugetlb: add a li... |
1258 1259 |
/* remove the page from active list */ list_del(&page->lru); |
a55164389 hugetlb: modular ... |
1260 1261 1262 |
update_and_free_page(h, page); h->surplus_huge_pages--; h->surplus_huge_pages_node[nid]--; |
7893d1d50 hugetlb: Try to g... |
1263 |
} else { |
5d3a551c2 mm: hugetlb: add ... |
1264 |
arch_clear_hugepage_flags(page); |
a55164389 hugetlb: modular ... |
1265 |
enqueue_huge_page(h, page); |
7893d1d50 hugetlb: Try to g... |
1266 |
} |
27a85ef1b [PATCH] hugepage:... |
1267 1268 |
spin_unlock(&hugetlb_lock); } |
3a43ea274 mm/hugetlb: defer... |
1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 |
/* * As free_huge_page() can be called from a non-task context, we have * to defer the actual freeing in a workqueue to prevent potential * hugetlb_lock deadlock. * * free_hpage_workfn() locklessly retrieves the linked list of pages to * be freed and frees them one-by-one. As the page->mapping pointer is * going to be cleared in __free_huge_page() anyway, it is reused as the * llist_node structure of a lockless linked list of huge pages to be freed. */ static LLIST_HEAD(hpage_freelist); static void free_hpage_workfn(struct work_struct *work) { struct llist_node *node; struct page *page; node = llist_del_all(&hpage_freelist); while (node) { page = container_of((struct address_space **)node, struct page, mapping); node = node->next; __free_huge_page(page); } } static DECLARE_WORK(free_hpage_work, free_hpage_workfn); void free_huge_page(struct page *page) { /* * Defer freeing if in non-task context to avoid hugetlb_lock deadlock. */ if (!in_task()) { /* * Only call schedule_work() if hpage_freelist is previously * empty. Otherwise, schedule_work() had been called but the * workfn hasn't retrieved the list yet. */ if (llist_add((struct llist_node *)&page->mapping, &hpage_freelist)) schedule_work(&free_hpage_work); return; } __free_huge_page(page); } |
a55164389 hugetlb: modular ... |
1316 |
static void prep_new_huge_page(struct hstate *h, struct page *page, int nid) |
b7ba30c67 hugetlb: factor o... |
1317 |
{ |
0edaecfab hugetlb: add a li... |
1318 |
INIT_LIST_HEAD(&page->lru); |
f1e61557f mm: pack compound... |
1319 |
set_compound_page_dtor(page, HUGETLB_PAGE_DTOR); |
b7ba30c67 hugetlb: factor o... |
1320 |
spin_lock(&hugetlb_lock); |
9dd540e23 hugetlb/cgroup: a... |
1321 |
set_hugetlb_cgroup(page, NULL); |
a55164389 hugetlb: modular ... |
1322 1323 |
h->nr_huge_pages++; h->nr_huge_pages_node[nid]++; |
b7ba30c67 hugetlb: factor o... |
1324 |
spin_unlock(&hugetlb_lock); |
b7ba30c67 hugetlb: factor o... |
1325 |
} |
d00181b96 mm: use 'unsigned... |
1326 |
static void prep_compound_gigantic_page(struct page *page, unsigned int order) |
20a0307c0 mm: introduce Pag... |
1327 1328 1329 1330 1331 1332 1333 |
{ 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); |
ef5a22be2 mm: hugetlb: init... |
1334 |
__ClearPageReserved(page); |
de09d31dd page-flags: defin... |
1335 |
__SetPageHead(page); |
20a0307c0 mm: introduce Pag... |
1336 |
for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) { |
ef5a22be2 mm: hugetlb: init... |
1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 |
/* * 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 thp: set compound... |
1350 |
set_page_count(p, 0); |
1d798ca3f mm: make compound... |
1351 |
set_compound_head(p, page); |
20a0307c0 mm: introduce Pag... |
1352 |
} |
b4330afbe mm/hugetlb: fix g... |
1353 |
atomic_set(compound_mapcount_ptr(page), -1); |
20a0307c0 mm: introduce Pag... |
1354 |
} |
7795912c2 mm: document Page... |
1355 1356 1357 1358 1359 |
/* * PageHuge() only returns true for hugetlbfs pages, but not for normal or * transparent huge pages. See the PageTransHuge() documentation for more * details. */ |
20a0307c0 mm: introduce Pag... |
1360 1361 |
int PageHuge(struct page *page) { |
20a0307c0 mm: introduce Pag... |
1362 1363 1364 1365 |
if (!PageCompound(page)) return 0; page = compound_head(page); |
f1e61557f mm: pack compound... |
1366 |
return page[1].compound_dtor == HUGETLB_PAGE_DTOR; |
20a0307c0 mm: introduce Pag... |
1367 |
} |
43131e141 HWPOISON, hugetlb... |
1368 |
EXPORT_SYMBOL_GPL(PageHuge); |
27c73ae75 mm: hugetlbfs: fi... |
1369 1370 1371 1372 1373 1374 |
/* * PageHeadHuge() only returns true for hugetlbfs head page, but not for * normal or transparent huge pages. */ int PageHeadHuge(struct page *page_head) { |
27c73ae75 mm: hugetlbfs: fi... |
1375 1376 |
if (!PageHead(page_head)) return 0; |
758f66a29 mm/hugetlb.c: sim... |
1377 |
return get_compound_page_dtor(page_head) == free_huge_page; |
27c73ae75 mm: hugetlbfs: fi... |
1378 |
} |
27c73ae75 mm: hugetlbfs: fi... |
1379 |
|
13d60f4b6 futex: Take hugep... |
1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 |
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; } |
0c397daea mm, hugetlb: furt... |
1396 |
static struct page *alloc_buddy_huge_page(struct hstate *h, |
f60858f9d hugetlbfs: don't ... |
1397 1398 |
gfp_t gfp_mask, int nid, nodemask_t *nmask, nodemask_t *node_alloc_noretry) |
1da177e4c Linux-2.6.12-rc2 |
1399 |
{ |
af0fb9df7 mm, hugetlb: unif... |
1400 |
int order = huge_page_order(h); |
1da177e4c Linux-2.6.12-rc2 |
1401 |
struct page *page; |
f60858f9d hugetlbfs: don't ... |
1402 |
bool alloc_try_hard = true; |
f96efd585 hugetlb: fix race... |
1403 |
|
f60858f9d hugetlbfs: don't ... |
1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 |
/* * By default we always try hard to allocate the page with * __GFP_RETRY_MAYFAIL flag. However, if we are allocating pages in * a loop (to adjust global huge page counts) and previous allocation * failed, do not continue to try hard on the same node. Use the * node_alloc_noretry bitmap to manage this state information. */ if (node_alloc_noretry && node_isset(nid, *node_alloc_noretry)) alloc_try_hard = false; gfp_mask |= __GFP_COMP|__GFP_NOWARN; if (alloc_try_hard) gfp_mask |= __GFP_RETRY_MAYFAIL; |
af0fb9df7 mm, hugetlb: unif... |
1416 1417 1418 1419 1420 1421 1422 |
if (nid == NUMA_NO_NODE) nid = numa_mem_id(); page = __alloc_pages_nodemask(gfp_mask, order, nid, nmask); if (page) __count_vm_event(HTLB_BUDDY_PGALLOC); else __count_vm_event(HTLB_BUDDY_PGALLOC_FAIL); |
63b4613c3 hugetlb: fix huge... |
1423 |
|
f60858f9d hugetlbfs: don't ... |
1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 |
/* * If we did not specify __GFP_RETRY_MAYFAIL, but still got a page this * indicates an overall state change. Clear bit so that we resume * normal 'try hard' allocations. */ if (node_alloc_noretry && page && !alloc_try_hard) node_clear(nid, *node_alloc_noretry); /* * If we tried hard to get a page but failed, set bit so that * subsequent attempts will not try as hard until there is an * overall state change. */ if (node_alloc_noretry && !page && alloc_try_hard) node_set(nid, *node_alloc_noretry); |
63b4613c3 hugetlb: fix huge... |
1439 1440 |
return page; } |
af0fb9df7 mm, hugetlb: unif... |
1441 |
/* |
0c397daea mm, hugetlb: furt... |
1442 1443 1444 1445 |
* Common helper to allocate a fresh hugetlb page. All specific allocators * should use this function to get new hugetlb pages */ static struct page *alloc_fresh_huge_page(struct hstate *h, |
f60858f9d hugetlbfs: don't ... |
1446 1447 |
gfp_t gfp_mask, int nid, nodemask_t *nmask, nodemask_t *node_alloc_noretry) |
0c397daea mm, hugetlb: furt... |
1448 1449 1450 1451 1452 1453 1454 |
{ struct page *page; if (hstate_is_gigantic(h)) page = alloc_gigantic_page(h, gfp_mask, nid, nmask); else page = alloc_buddy_huge_page(h, gfp_mask, |
f60858f9d hugetlbfs: don't ... |
1455 |
nid, nmask, node_alloc_noretry); |
0c397daea mm, hugetlb: furt... |
1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 |
if (!page) return NULL; if (hstate_is_gigantic(h)) prep_compound_gigantic_page(page, huge_page_order(h)); prep_new_huge_page(h, page, page_to_nid(page)); return page; } /* |
af0fb9df7 mm, hugetlb: unif... |
1467 1468 1469 |
* Allocates a fresh page to the hugetlb allocator pool in the node interleaved * manner. */ |
f60858f9d hugetlbfs: don't ... |
1470 1471 |
static int alloc_pool_huge_page(struct hstate *h, nodemask_t *nodes_allowed, nodemask_t *node_alloc_noretry) |
b22610268 mm, hugetlb: fix ... |
1472 1473 1474 |
{ struct page *page; int nr_nodes, node; |
af0fb9df7 mm, hugetlb: unif... |
1475 |
gfp_t gfp_mask = htlb_alloc_mask(h) | __GFP_THISNODE; |
b22610268 mm, hugetlb: fix ... |
1476 1477 |
for_each_node_mask_to_alloc(h, nr_nodes, node, nodes_allowed) { |
f60858f9d hugetlbfs: don't ... |
1478 1479 |
page = alloc_fresh_huge_page(h, gfp_mask, node, nodes_allowed, node_alloc_noretry); |
af0fb9df7 mm, hugetlb: unif... |
1480 |
if (page) |
b22610268 mm, hugetlb: fix ... |
1481 |
break; |
b22610268 mm, hugetlb: fix ... |
1482 |
} |
af0fb9df7 mm, hugetlb: unif... |
1483 1484 |
if (!page) return 0; |
b22610268 mm, hugetlb: fix ... |
1485 |
|
af0fb9df7 mm, hugetlb: unif... |
1486 1487 1488 |
put_page(page); /* free it into the hugepage allocator */ return 1; |
b22610268 mm, hugetlb: fix ... |
1489 |
} |
e8c5c8249 hugetlb: balance ... |
1490 1491 1492 1493 1494 1495 |
/* * 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 hugetlb: add node... |
1496 1497 |
static int free_pool_huge_page(struct hstate *h, nodemask_t *nodes_allowed, bool acct_surplus) |
e8c5c8249 hugetlb: balance ... |
1498 |
{ |
b22610268 mm, hugetlb: fix ... |
1499 |
int nr_nodes, node; |
e8c5c8249 hugetlb: balance ... |
1500 |
int ret = 0; |
b22610268 mm, hugetlb: fix ... |
1501 |
for_each_node_mask_to_free(h, nr_nodes, node, nodes_allowed) { |
685f34570 hugetlb: use free... |
1502 1503 1504 1505 |
/* * If we're returning unused surplus pages, only examine * nodes with surplus pages. */ |
b22610268 mm, hugetlb: fix ... |
1506 1507 |
if ((!acct_surplus || h->surplus_huge_pages_node[node]) && !list_empty(&h->hugepage_freelists[node])) { |
e8c5c8249 hugetlb: balance ... |
1508 |
struct page *page = |
b22610268 mm, hugetlb: fix ... |
1509 |
list_entry(h->hugepage_freelists[node].next, |
e8c5c8249 hugetlb: balance ... |
1510 1511 1512 |
struct page, lru); list_del(&page->lru); h->free_huge_pages--; |
b22610268 mm, hugetlb: fix ... |
1513 |
h->free_huge_pages_node[node]--; |
685f34570 hugetlb: use free... |
1514 1515 |
if (acct_surplus) { h->surplus_huge_pages--; |
b22610268 mm, hugetlb: fix ... |
1516 |
h->surplus_huge_pages_node[node]--; |
685f34570 hugetlb: use free... |
1517 |
} |
e8c5c8249 hugetlb: balance ... |
1518 1519 |
update_and_free_page(h, page); ret = 1; |
9a76db099 hugetlb: rework h... |
1520 |
break; |
e8c5c8249 hugetlb: balance ... |
1521 |
} |
b22610268 mm, hugetlb: fix ... |
1522 |
} |
e8c5c8249 hugetlb: balance ... |
1523 1524 1525 |
return ret; } |
c8721bbbd mm: memory-hotplu... |
1526 1527 |
/* * Dissolve a given free hugepage into free buddy pages. This function does |
faf53def3 mm: hugetlb: soft... |
1528 1529 1530 1531 1532 1533 1534 |
* nothing for in-use hugepages and non-hugepages. * This function returns values like below: * * -EBUSY: failed to dissolved free hugepages or the hugepage is in-use * (allocated or reserved.) * 0: successfully dissolved free hugepages or the page is not a * hugepage (considered as already dissolved) |
c8721bbbd mm: memory-hotplu... |
1535 |
*/ |
c3114a84f mm: hugetlb: soft... |
1536 |
int dissolve_free_huge_page(struct page *page) |
c8721bbbd mm: memory-hotplu... |
1537 |
{ |
6bc9b5643 mm: fix race on s... |
1538 |
int rc = -EBUSY; |
082d5b6b6 mm/hugetlb: check... |
1539 |
|
faf53def3 mm: hugetlb: soft... |
1540 1541 1542 |
/* Not to disrupt normal path by vainly holding hugetlb_lock */ if (!PageHuge(page)) return 0; |
c8721bbbd mm: memory-hotplu... |
1543 |
spin_lock(&hugetlb_lock); |
faf53def3 mm: hugetlb: soft... |
1544 1545 1546 1547 1548 1549 |
if (!PageHuge(page)) { rc = 0; goto out; } if (!page_count(page)) { |
2247bb335 mm/hugetlb: fix m... |
1550 1551 1552 |
struct page *head = compound_head(page); struct hstate *h = page_hstate(head); int nid = page_to_nid(head); |
6bc9b5643 mm: fix race on s... |
1553 |
if (h->free_huge_pages - h->resv_huge_pages == 0) |
082d5b6b6 mm/hugetlb: check... |
1554 |
goto out; |
c3114a84f mm: hugetlb: soft... |
1555 1556 1557 1558 1559 1560 1561 1562 |
/* * Move PageHWPoison flag from head page to the raw error page, * which makes any subpages rather than the error page reusable. */ if (PageHWPoison(head) && page != head) { SetPageHWPoison(page); ClearPageHWPoison(head); } |
2247bb335 mm/hugetlb: fix m... |
1563 |
list_del(&head->lru); |
c8721bbbd mm: memory-hotplu... |
1564 1565 |
h->free_huge_pages--; h->free_huge_pages_node[nid]--; |
c1470b33b mm/hugetlb: fix i... |
1566 |
h->max_huge_pages--; |
2247bb335 mm/hugetlb: fix m... |
1567 |
update_and_free_page(h, head); |
6bc9b5643 mm: fix race on s... |
1568 |
rc = 0; |
c8721bbbd mm: memory-hotplu... |
1569 |
} |
082d5b6b6 mm/hugetlb: check... |
1570 |
out: |
c8721bbbd mm: memory-hotplu... |
1571 |
spin_unlock(&hugetlb_lock); |
082d5b6b6 mm/hugetlb: check... |
1572 |
return rc; |
c8721bbbd mm: memory-hotplu... |
1573 1574 1575 1576 1577 |
} /* * Dissolve free hugepages in a given pfn range. Used by memory hotplug to * make specified memory blocks removable from the system. |
2247bb335 mm/hugetlb: fix m... |
1578 1579 |
* Note that this will dissolve a free gigantic hugepage completely, if any * part of it lies within the given range. |
082d5b6b6 mm/hugetlb: check... |
1580 1581 |
* Also note that if dissolve_free_huge_page() returns with an error, all * free hugepages that were dissolved before that error are lost. |
c8721bbbd mm: memory-hotplu... |
1582 |
*/ |
082d5b6b6 mm/hugetlb: check... |
1583 |
int dissolve_free_huge_pages(unsigned long start_pfn, unsigned long end_pfn) |
c8721bbbd mm: memory-hotplu... |
1584 |
{ |
c8721bbbd mm: memory-hotplu... |
1585 |
unsigned long pfn; |
eb03aa008 mm/hugetlb: impro... |
1586 |
struct page *page; |
082d5b6b6 mm/hugetlb: check... |
1587 |
int rc = 0; |
c8721bbbd mm: memory-hotplu... |
1588 |
|
d01776393 mm: fix potential... |
1589 |
if (!hugepages_supported()) |
082d5b6b6 mm/hugetlb: check... |
1590 |
return rc; |
d01776393 mm: fix potential... |
1591 |
|
eb03aa008 mm/hugetlb: impro... |
1592 1593 |
for (pfn = start_pfn; pfn < end_pfn; pfn += 1 << minimum_order) { page = pfn_to_page(pfn); |
faf53def3 mm: hugetlb: soft... |
1594 1595 1596 |
rc = dissolve_free_huge_page(page); if (rc) break; |
eb03aa008 mm/hugetlb: impro... |
1597 |
} |
082d5b6b6 mm/hugetlb: check... |
1598 1599 |
return rc; |
c8721bbbd mm: memory-hotplu... |
1600 |
} |
ab5ac90ae mm, hugetlb: do n... |
1601 1602 1603 |
/* * Allocates a fresh surplus page from the page allocator. */ |
0c397daea mm, hugetlb: furt... |
1604 |
static struct page *alloc_surplus_huge_page(struct hstate *h, gfp_t gfp_mask, |
aaf14e40a mm, hugetlb: uncl... |
1605 |
int nid, nodemask_t *nmask) |
7893d1d50 hugetlb: Try to g... |
1606 |
{ |
9980d744a mm, hugetlb: get ... |
1607 |
struct page *page = NULL; |
7893d1d50 hugetlb: Try to g... |
1608 |
|
bae7f4ae1 hugetlb: add hsta... |
1609 |
if (hstate_is_gigantic(h)) |
aa888a749 hugetlb: support ... |
1610 |
return NULL; |
d1c3fb1f8 hugetlb: introduc... |
1611 |
spin_lock(&hugetlb_lock); |
9980d744a mm, hugetlb: get ... |
1612 1613 |
if (h->surplus_huge_pages >= h->nr_overcommit_huge_pages) goto out_unlock; |
d1c3fb1f8 hugetlb: introduc... |
1614 |
spin_unlock(&hugetlb_lock); |
f60858f9d hugetlbfs: don't ... |
1615 |
page = alloc_fresh_huge_page(h, gfp_mask, nid, nmask, NULL); |
9980d744a mm, hugetlb: get ... |
1616 |
if (!page) |
0c397daea mm, hugetlb: furt... |
1617 |
return NULL; |
d1c3fb1f8 hugetlb: introduc... |
1618 1619 |
spin_lock(&hugetlb_lock); |
9980d744a mm, hugetlb: get ... |
1620 1621 1622 1623 1624 1625 1626 1627 1628 |
/* * We could have raced with the pool size change. * Double check that and simply deallocate the new page * if we would end up overcommiting the surpluses. Abuse * temporary page to workaround the nasty free_huge_page * codeflow */ if (h->surplus_huge_pages >= h->nr_overcommit_huge_pages) { SetPageHugeTemporary(page); |
2bf753e64 mm/hugetlb.c: don... |
1629 |
spin_unlock(&hugetlb_lock); |
9980d744a mm, hugetlb: get ... |
1630 |
put_page(page); |
2bf753e64 mm/hugetlb.c: don... |
1631 |
return NULL; |
9980d744a mm, hugetlb: get ... |
1632 |
} else { |
9980d744a mm, hugetlb: get ... |
1633 |
h->surplus_huge_pages++; |
4704dea36 hugetlb: fix surp... |
1634 |
h->surplus_huge_pages_node[page_to_nid(page)]++; |
7893d1d50 hugetlb: Try to g... |
1635 |
} |
9980d744a mm, hugetlb: get ... |
1636 1637 |
out_unlock: |
d1c3fb1f8 hugetlb: introduc... |
1638 |
spin_unlock(&hugetlb_lock); |
7893d1d50 hugetlb: Try to g... |
1639 1640 1641 |
return page; } |
9a4e9f3b2 mm: update get_us... |
1642 1643 |
struct page *alloc_migrate_huge_page(struct hstate *h, gfp_t gfp_mask, int nid, nodemask_t *nmask) |
ab5ac90ae mm, hugetlb: do n... |
1644 1645 1646 1647 1648 |
{ struct page *page; if (hstate_is_gigantic(h)) return NULL; |
f60858f9d hugetlbfs: don't ... |
1649 |
page = alloc_fresh_huge_page(h, gfp_mask, nid, nmask, NULL); |
ab5ac90ae mm, hugetlb: do n... |
1650 1651 1652 1653 1654 1655 1656 |
if (!page) return NULL; /* * We do not account these pages as surplus because they are only * temporary and will be released properly on the last reference */ |
ab5ac90ae mm, hugetlb: do n... |
1657 1658 1659 1660 |
SetPageHugeTemporary(page); return page; } |
e4e574b76 hugetlb: Try to g... |
1661 |
/* |
099730d67 mm, hugetlb: use ... |
1662 1663 |
* Use the VMA's mpolicy to allocate a huge page from the buddy. */ |
e0ec90ee7 mm, hugetlbfs: op... |
1664 |
static |
0c397daea mm, hugetlb: furt... |
1665 |
struct page *alloc_buddy_huge_page_with_mpol(struct hstate *h, |
099730d67 mm, hugetlb: use ... |
1666 1667 |
struct vm_area_struct *vma, unsigned long addr) { |
aaf14e40a mm, hugetlb: uncl... |
1668 1669 1670 1671 1672 1673 1674 |
struct page *page; struct mempolicy *mpol; gfp_t gfp_mask = htlb_alloc_mask(h); int nid; nodemask_t *nodemask; nid = huge_node(vma, addr, gfp_mask, &mpol, &nodemask); |
0c397daea mm, hugetlb: furt... |
1675 |
page = alloc_surplus_huge_page(h, gfp_mask, nid, nodemask); |
aaf14e40a mm, hugetlb: uncl... |
1676 1677 1678 |
mpol_cond_put(mpol); return page; |
099730d67 mm, hugetlb: use ... |
1679 |
} |
ab5ac90ae mm, hugetlb: do n... |
1680 |
/* page migration callback function */ |
bf50bab2b hugetlb: add allo... |
1681 1682 |
struct page *alloc_huge_page_node(struct hstate *h, int nid) { |
aaf14e40a mm, hugetlb: uncl... |
1683 |
gfp_t gfp_mask = htlb_alloc_mask(h); |
4ef918480 mm, hugetlb: prot... |
1684 |
struct page *page = NULL; |
bf50bab2b hugetlb: add allo... |
1685 |
|
aaf14e40a mm, hugetlb: uncl... |
1686 1687 |
if (nid != NUMA_NO_NODE) gfp_mask |= __GFP_THISNODE; |
bf50bab2b hugetlb: add allo... |
1688 |
spin_lock(&hugetlb_lock); |
4ef918480 mm, hugetlb: prot... |
1689 |
if (h->free_huge_pages - h->resv_huge_pages > 0) |
3e59fcb0e hugetlb: add supp... |
1690 |
page = dequeue_huge_page_nodemask(h, gfp_mask, nid, NULL); |
bf50bab2b hugetlb: add allo... |
1691 |
spin_unlock(&hugetlb_lock); |
94ae8ba71 hugetlb/cgroup: a... |
1692 |
if (!page) |
0c397daea mm, hugetlb: furt... |
1693 |
page = alloc_migrate_huge_page(h, gfp_mask, nid, NULL); |
bf50bab2b hugetlb: add allo... |
1694 1695 1696 |
return page; } |
ab5ac90ae mm, hugetlb: do n... |
1697 |
/* page migration callback function */ |
3e59fcb0e hugetlb: add supp... |
1698 1699 |
struct page *alloc_huge_page_nodemask(struct hstate *h, int preferred_nid, nodemask_t *nmask) |
4db9b2efe hugetlb, memory_h... |
1700 |
{ |
aaf14e40a mm, hugetlb: uncl... |
1701 |
gfp_t gfp_mask = htlb_alloc_mask(h); |
4db9b2efe hugetlb, memory_h... |
1702 1703 1704 |
spin_lock(&hugetlb_lock); if (h->free_huge_pages - h->resv_huge_pages > 0) { |
3e59fcb0e hugetlb: add supp... |
1705 1706 1707 1708 1709 1710 |
struct page *page; page = dequeue_huge_page_nodemask(h, gfp_mask, preferred_nid, nmask); if (page) { spin_unlock(&hugetlb_lock); return page; |
4db9b2efe hugetlb, memory_h... |
1711 1712 1713 |
} } spin_unlock(&hugetlb_lock); |
4db9b2efe hugetlb, memory_h... |
1714 |
|
0c397daea mm, hugetlb: furt... |
1715 |
return alloc_migrate_huge_page(h, gfp_mask, preferred_nid, nmask); |
4db9b2efe hugetlb, memory_h... |
1716 |
} |
ebd637235 hugetlb, mempolic... |
1717 |
/* mempolicy aware migration callback */ |
389c8178d hugetlb, mbind: f... |
1718 1719 |
struct page *alloc_huge_page_vma(struct hstate *h, struct vm_area_struct *vma, unsigned long address) |
ebd637235 hugetlb, mempolic... |
1720 1721 1722 1723 |
{ struct mempolicy *mpol; nodemask_t *nodemask; struct page *page; |
ebd637235 hugetlb, mempolic... |
1724 1725 |
gfp_t gfp_mask; int node; |
ebd637235 hugetlb, mempolic... |
1726 1727 1728 1729 1730 1731 1732 |
gfp_mask = htlb_alloc_mask(h); node = huge_node(vma, address, gfp_mask, &mpol, &nodemask); page = alloc_huge_page_nodemask(h, node, nodemask); mpol_cond_put(mpol); return page; } |
bf50bab2b hugetlb: add allo... |
1733 |
/* |
25985edce Fix common misspe... |
1734 |
* Increase the hugetlb pool such that it can accommodate a reservation |
e4e574b76 hugetlb: Try to g... |
1735 1736 |
* of size 'delta'. */ |
a55164389 hugetlb: modular ... |
1737 |
static int gather_surplus_pages(struct hstate *h, int delta) |
e4e574b76 hugetlb: Try to g... |
1738 1739 1740 1741 1742 |
{ struct list_head surplus_list; struct page *page, *tmp; int ret, i; int needed, allocated; |
28073b02b mm: hugetlb: defe... |
1743 |
bool alloc_ok = true; |
e4e574b76 hugetlb: Try to g... |
1744 |
|
a55164389 hugetlb: modular ... |
1745 |
needed = (h->resv_huge_pages + delta) - h->free_huge_pages; |
ac09b3a15 hugetlb: close a ... |
1746 |
if (needed <= 0) { |
a55164389 hugetlb: modular ... |
1747 |
h->resv_huge_pages += delta; |
e4e574b76 hugetlb: Try to g... |
1748 |
return 0; |
ac09b3a15 hugetlb: close a ... |
1749 |
} |
e4e574b76 hugetlb: Try to g... |
1750 1751 1752 1753 1754 1755 1756 1757 |
allocated = 0; INIT_LIST_HEAD(&surplus_list); ret = -ENOMEM; retry: spin_unlock(&hugetlb_lock); for (i = 0; i < needed; i++) { |
0c397daea mm, hugetlb: furt... |
1758 |
page = alloc_surplus_huge_page(h, htlb_alloc_mask(h), |
aaf14e40a mm, hugetlb: uncl... |
1759 |
NUMA_NO_NODE, NULL); |
28073b02b mm: hugetlb: defe... |
1760 1761 1762 1763 |
if (!page) { alloc_ok = false; break; } |
e4e574b76 hugetlb: Try to g... |
1764 |
list_add(&page->lru, &surplus_list); |
69ed779a1 mm, hugetlb: sche... |
1765 |
cond_resched(); |
e4e574b76 hugetlb: Try to g... |
1766 |
} |
28073b02b mm: hugetlb: defe... |
1767 |
allocated += i; |
e4e574b76 hugetlb: Try to g... |
1768 1769 1770 1771 1772 1773 |
/* * 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 hugetlb: modular ... |
1774 1775 |
needed = (h->resv_huge_pages + delta) - (h->free_huge_pages + allocated); |
28073b02b mm: hugetlb: defe... |
1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 |
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 hugetlb: Try to g... |
1786 1787 |
/* * The surplus_list now contains _at_least_ the number of extra pages |
25985edce Fix common misspe... |
1788 |
* needed to accommodate the reservation. Add the appropriate number |
e4e574b76 hugetlb: Try to g... |
1789 |
* of pages to the hugetlb pool and free the extras back to the buddy |
ac09b3a15 hugetlb: close a ... |
1790 1791 1792 |
* 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 hugetlb: Try to g... |
1793 1794 |
*/ needed += allocated; |
a55164389 hugetlb: modular ... |
1795 |
h->resv_huge_pages += delta; |
e4e574b76 hugetlb: Try to g... |
1796 |
ret = 0; |
a9869b837 hugetlb: move ref... |
1797 |
|
19fc3f0ac hugetlb: decrease... |
1798 |
/* Free the needed pages to the hugetlb pool */ |
e4e574b76 hugetlb: Try to g... |
1799 |
list_for_each_entry_safe(page, tmp, &surplus_list, lru) { |
19fc3f0ac hugetlb: decrease... |
1800 1801 |
if ((--needed) < 0) break; |
a9869b837 hugetlb: move ref... |
1802 1803 1804 1805 1806 |
/* * This page is now managed by the hugetlb allocator and has * no users -- drop the buddy allocator's reference. */ put_page_testzero(page); |
309381fea mm: dump page whe... |
1807 |
VM_BUG_ON_PAGE(page_count(page), page); |
a55164389 hugetlb: modular ... |
1808 |
enqueue_huge_page(h, page); |
19fc3f0ac hugetlb: decrease... |
1809 |
} |
28073b02b mm: hugetlb: defe... |
1810 |
free: |
b0365c8d0 mm: hugetlb: fix ... |
1811 |
spin_unlock(&hugetlb_lock); |
19fc3f0ac hugetlb: decrease... |
1812 1813 |
/* Free unnecessary surplus pages to the buddy allocator */ |
c0d934ba2 mm, hugetlb: remo... |
1814 1815 |
list_for_each_entry_safe(page, tmp, &surplus_list, lru) put_page(page); |
a9869b837 hugetlb: move ref... |
1816 |
spin_lock(&hugetlb_lock); |
e4e574b76 hugetlb: Try to g... |
1817 1818 1819 1820 1821 |
return ret; } /* |
e5bbc8a6c mm/hugetlb.c: fix... |
1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 |
* This routine has two main purposes: * 1) Decrement the reservation count (resv_huge_pages) by the value passed * in unused_resv_pages. This corresponds to the prior adjustments made * to the associated reservation map. * 2) Free any unused surplus pages that may have been allocated to satisfy * the reservation. As many as unused_resv_pages may be freed. * * Called with hugetlb_lock held. However, the lock could be dropped (and * reacquired) during calls to cond_resched_lock. Whenever dropping the lock, * we must make sure nobody else can claim pages we are in the process of * freeing. Do this by ensuring resv_huge_page always is greater than the * number of huge pages we plan to free when dropping the lock. |
e4e574b76 hugetlb: Try to g... |
1834 |
*/ |
a55164389 hugetlb: modular ... |
1835 1836 |
static void return_unused_surplus_pages(struct hstate *h, unsigned long unused_resv_pages) |
e4e574b76 hugetlb: Try to g... |
1837 |
{ |
e4e574b76 hugetlb: Try to g... |
1838 |
unsigned long nr_pages; |
aa888a749 hugetlb: support ... |
1839 |
/* Cannot return gigantic pages currently */ |
bae7f4ae1 hugetlb: add hsta... |
1840 |
if (hstate_is_gigantic(h)) |
e5bbc8a6c mm/hugetlb.c: fix... |
1841 |
goto out; |
aa888a749 hugetlb: support ... |
1842 |
|
e5bbc8a6c mm/hugetlb.c: fix... |
1843 1844 1845 1846 |
/* * Part (or even all) of the reservation could have been backed * by pre-allocated pages. Only free surplus pages. */ |
a55164389 hugetlb: modular ... |
1847 |
nr_pages = min(unused_resv_pages, h->surplus_huge_pages); |
e4e574b76 hugetlb: Try to g... |
1848 |
|
685f34570 hugetlb: use free... |
1849 1850 |
/* * We want to release as many surplus pages as possible, spread |
9b5e5d0fd hugetlb: use only... |
1851 1852 1853 1854 1855 |
* 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. |
e5bbc8a6c mm/hugetlb.c: fix... |
1856 1857 1858 1859 |
* * Note that we decrement resv_huge_pages as we free the pages. If * we drop the lock, resv_huge_pages will still be sufficiently large * to cover subsequent pages we may free. |
685f34570 hugetlb: use free... |
1860 1861 |
*/ while (nr_pages--) { |
e5bbc8a6c mm/hugetlb.c: fix... |
1862 1863 |
h->resv_huge_pages--; unused_resv_pages--; |
8cebfcd07 hugetlb: use N_ME... |
1864 |
if (!free_pool_huge_page(h, &node_states[N_MEMORY], 1)) |
e5bbc8a6c mm/hugetlb.c: fix... |
1865 |
goto out; |
7848a4bf5 mm/hugetlb.c: add... |
1866 |
cond_resched_lock(&hugetlb_lock); |
e4e574b76 hugetlb: Try to g... |
1867 |
} |
e5bbc8a6c mm/hugetlb.c: fix... |
1868 1869 1870 1871 |
out: /* Fully uncommit the reservation */ h->resv_huge_pages -= unused_resv_pages; |
e4e574b76 hugetlb: Try to g... |
1872 |
} |
5e9113731 mm/hugetlb: add c... |
1873 |
|
c37f9fb11 hugetlb: allow hu... |
1874 |
/* |
feba16e25 mm/hugetlb: add r... |
1875 |
* vma_needs_reservation, vma_commit_reservation and vma_end_reservation |
5e9113731 mm/hugetlb: add c... |
1876 |
* are used by the huge page allocation routines to manage reservations. |
cf3ad20bf mm/hugetlb: compu... |
1877 1878 1879 1880 1881 1882 |
* * vma_needs_reservation is called to determine if the huge page at addr * within the vma has an associated reservation. If a reservation is * needed, the value 1 is returned. The caller is then responsible for * managing the global reservation and subpool usage counts. After * the huge page has been allocated, vma_commit_reservation is called |
feba16e25 mm/hugetlb: add r... |
1883 1884 1885 |
* to add the page to the reservation map. If the page allocation fails, * the reservation must be ended instead of committed. vma_end_reservation * is called in such cases. |
cf3ad20bf mm/hugetlb: compu... |
1886 1887 1888 1889 1890 1891 |
* * In the normal case, vma_commit_reservation returns the same value * as the preceding vma_needs_reservation call. The only time this * is not the case is if a reserve map was changed between calls. It * is the responsibility of the caller to notice the difference and * take appropriate action. |
96b96a96d mm/hugetlb: fix h... |
1892 1893 1894 1895 1896 |
* * vma_add_reservation is used in error paths where a reservation must * be restored when a newly allocated huge page must be freed. It is * to be called after calling vma_needs_reservation to determine if a * reservation exists. |
c37f9fb11 hugetlb: allow hu... |
1897 |
*/ |
5e9113731 mm/hugetlb: add c... |
1898 1899 1900 |
enum vma_resv_mode { VMA_NEEDS_RESV, VMA_COMMIT_RESV, |
feba16e25 mm/hugetlb: add r... |
1901 |
VMA_END_RESV, |
96b96a96d mm/hugetlb: fix h... |
1902 |
VMA_ADD_RESV, |
5e9113731 mm/hugetlb: add c... |
1903 |
}; |
cf3ad20bf mm/hugetlb: compu... |
1904 1905 |
static long __vma_reservation_common(struct hstate *h, struct vm_area_struct *vma, unsigned long addr, |
5e9113731 mm/hugetlb: add c... |
1906 |
enum vma_resv_mode mode) |
c37f9fb11 hugetlb: allow hu... |
1907 |
{ |
4e35f4838 mm, hugetlb: use ... |
1908 1909 |
struct resv_map *resv; pgoff_t idx; |
cf3ad20bf mm/hugetlb: compu... |
1910 |
long ret; |
c37f9fb11 hugetlb: allow hu... |
1911 |
|
4e35f4838 mm, hugetlb: use ... |
1912 1913 |
resv = vma_resv_map(vma); if (!resv) |
84afd99b8 hugetlb reservati... |
1914 |
return 1; |
c37f9fb11 hugetlb: allow hu... |
1915 |
|
4e35f4838 mm, hugetlb: use ... |
1916 |
idx = vma_hugecache_offset(h, vma, addr); |
5e9113731 mm/hugetlb: add c... |
1917 1918 |
switch (mode) { case VMA_NEEDS_RESV: |
cf3ad20bf mm/hugetlb: compu... |
1919 |
ret = region_chg(resv, idx, idx + 1); |
5e9113731 mm/hugetlb: add c... |
1920 1921 1922 1923 |
break; case VMA_COMMIT_RESV: ret = region_add(resv, idx, idx + 1); break; |
feba16e25 mm/hugetlb: add r... |
1924 |
case VMA_END_RESV: |
5e9113731 mm/hugetlb: add c... |
1925 1926 1927 |
region_abort(resv, idx, idx + 1); ret = 0; break; |
96b96a96d mm/hugetlb: fix h... |
1928 1929 1930 1931 1932 1933 1934 1935 |
case VMA_ADD_RESV: if (vma->vm_flags & VM_MAYSHARE) ret = region_add(resv, idx, idx + 1); else { region_abort(resv, idx, idx + 1); ret = region_del(resv, idx, idx + 1); } break; |
5e9113731 mm/hugetlb: add c... |
1936 1937 1938 |
default: BUG(); } |
84afd99b8 hugetlb reservati... |
1939 |
|
4e35f4838 mm, hugetlb: use ... |
1940 |
if (vma->vm_flags & VM_MAYSHARE) |
cf3ad20bf mm/hugetlb: compu... |
1941 |
return ret; |
67961f9db mm/hugetlb: fix h... |
1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 |
else if (is_vma_resv_set(vma, HPAGE_RESV_OWNER) && ret >= 0) { /* * In most cases, reserves always exist for private mappings. * However, a file associated with mapping could have been * hole punched or truncated after reserves were consumed. * As subsequent fault on such a range will not use reserves. * Subtle - The reserve map for private mappings has the * opposite meaning than that of shared mappings. If NO * entry is in the reserve map, it means a reservation exists. * If an entry exists in the reserve map, it means the * reservation has already been consumed. As a result, the * return value of this routine is the opposite of the * value returned from reserve map manipulation routines above. */ if (ret) return 0; else return 1; } |
4e35f4838 mm, hugetlb: use ... |
1961 |
else |
cf3ad20bf mm/hugetlb: compu... |
1962 |
return ret < 0 ? ret : 0; |
c37f9fb11 hugetlb: allow hu... |
1963 |
} |
cf3ad20bf mm/hugetlb: compu... |
1964 1965 |
static long vma_needs_reservation(struct hstate *h, |
a55164389 hugetlb: modular ... |
1966 |
struct vm_area_struct *vma, unsigned long addr) |
c37f9fb11 hugetlb: allow hu... |
1967 |
{ |
5e9113731 mm/hugetlb: add c... |
1968 |
return __vma_reservation_common(h, vma, addr, VMA_NEEDS_RESV); |
cf3ad20bf mm/hugetlb: compu... |
1969 |
} |
84afd99b8 hugetlb reservati... |
1970 |
|
cf3ad20bf mm/hugetlb: compu... |
1971 1972 1973 |
static long vma_commit_reservation(struct hstate *h, struct vm_area_struct *vma, unsigned long addr) { |
5e9113731 mm/hugetlb: add c... |
1974 1975 |
return __vma_reservation_common(h, vma, addr, VMA_COMMIT_RESV); } |
feba16e25 mm/hugetlb: add r... |
1976 |
static void vma_end_reservation(struct hstate *h, |
5e9113731 mm/hugetlb: add c... |
1977 1978 |
struct vm_area_struct *vma, unsigned long addr) { |
feba16e25 mm/hugetlb: add r... |
1979 |
(void)__vma_reservation_common(h, vma, addr, VMA_END_RESV); |
c37f9fb11 hugetlb: allow hu... |
1980 |
} |
96b96a96d mm/hugetlb: fix h... |
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 |
static long vma_add_reservation(struct hstate *h, struct vm_area_struct *vma, unsigned long addr) { return __vma_reservation_common(h, vma, addr, VMA_ADD_RESV); } /* * This routine is called to restore a reservation on error paths. In the * specific error paths, a huge page was allocated (via alloc_huge_page) * and is about to be freed. If a reservation for the page existed, * alloc_huge_page would have consumed the reservation and set PagePrivate * in the newly allocated page. When the page is freed via free_huge_page, * the global reservation count will be incremented if PagePrivate is set. * However, free_huge_page can not adjust the reserve map. Adjust the * reserve map here to be consistent with global reserve count adjustments * to be made by free_huge_page. */ static void restore_reserve_on_error(struct hstate *h, struct vm_area_struct *vma, unsigned long address, struct page *page) { if (unlikely(PagePrivate(page))) { long rc = vma_needs_reservation(h, vma, address); if (unlikely(rc < 0)) { /* * Rare out of memory condition in reserve map * manipulation. Clear PagePrivate so that * global reserve count will not be incremented * by free_huge_page. This will make it appear * as though the reservation for this page was * consumed. This may prevent the task from * faulting in the page at a later time. This * is better than inconsistent global huge page * accounting of reserve counts. */ ClearPagePrivate(page); } else if (rc) { rc = vma_add_reservation(h, vma, address); if (unlikely(rc < 0)) /* * See above comment about rare out of * memory condition. */ ClearPagePrivate(page); } else vma_end_reservation(h, vma, address); } } |
70c3547e3 hugetlbfs: add hu... |
2030 |
struct page *alloc_huge_page(struct vm_area_struct *vma, |
04f2cbe35 hugetlb: guarante... |
2031 |
unsigned long addr, int avoid_reserve) |
1da177e4c Linux-2.6.12-rc2 |
2032 |
{ |
90481622d hugepages: fix us... |
2033 |
struct hugepage_subpool *spool = subpool_vma(vma); |
a55164389 hugetlb: modular ... |
2034 |
struct hstate *h = hstate_vma(vma); |
348ea204c hugetlb: split al... |
2035 |
struct page *page; |
d85f69b0b mm/hugetlb: alloc... |
2036 2037 |
long map_chg, map_commit; long gbl_chg; |
6d76dcf40 hugetlb/cgroup: a... |
2038 2039 |
int ret, idx; struct hugetlb_cgroup *h_cg; |
a1e78772d hugetlb: reserve ... |
2040 |
|
6d76dcf40 hugetlb/cgroup: a... |
2041 |
idx = hstate_index(h); |
a1e78772d hugetlb: reserve ... |
2042 |
/* |
d85f69b0b mm/hugetlb: alloc... |
2043 2044 2045 |
* Examine the region/reserve map to determine if the process * has a reservation for the page to be allocated. A return * code of zero indicates a reservation exists (no change). |
a1e78772d hugetlb: reserve ... |
2046 |
*/ |
d85f69b0b mm/hugetlb: alloc... |
2047 2048 |
map_chg = gbl_chg = vma_needs_reservation(h, vma, addr); if (map_chg < 0) |
76dcee75c hugetlb: don't us... |
2049 |
return ERR_PTR(-ENOMEM); |
d85f69b0b mm/hugetlb: alloc... |
2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 |
/* * Processes that did not create the mapping will have no * reserves as indicated by the region/reserve map. Check * that the allocation will not exceed the subpool limit. * Allocations for MAP_NORESERVE mappings also need to be * checked against any subpool limit. */ if (map_chg || avoid_reserve) { gbl_chg = hugepage_subpool_get_pages(spool, 1); if (gbl_chg < 0) { |
feba16e25 mm/hugetlb: add r... |
2061 |
vma_end_reservation(h, vma, addr); |
76dcee75c hugetlb: don't us... |
2062 |
return ERR_PTR(-ENOSPC); |
5e9113731 mm/hugetlb: add c... |
2063 |
} |
1da177e4c Linux-2.6.12-rc2 |
2064 |
|
d85f69b0b mm/hugetlb: alloc... |
2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 |
/* * Even though there was no reservation in the region/reserve * map, there could be reservations associated with the * subpool that can be used. This would be indicated if the * return value of hugepage_subpool_get_pages() is zero. * However, if avoid_reserve is specified we still avoid even * the subpool reservations. */ if (avoid_reserve) gbl_chg = 1; } |
6d76dcf40 hugetlb/cgroup: a... |
2076 |
ret = hugetlb_cgroup_charge_cgroup(idx, pages_per_huge_page(h), &h_cg); |
8f34af6f9 mm, hugetlb: move... |
2077 2078 |
if (ret) goto out_subpool_put; |
1da177e4c Linux-2.6.12-rc2 |
2079 |
spin_lock(&hugetlb_lock); |
d85f69b0b mm/hugetlb: alloc... |
2080 2081 2082 2083 2084 2085 |
/* * glb_chg is passed to indicate whether or not a page must be taken * from the global free pool (global change). gbl_chg == 0 indicates * a reservation exists for the allocation. */ page = dequeue_huge_page_vma(h, vma, addr, avoid_reserve, gbl_chg); |
81a6fcae3 mm, hugetlb: clea... |
2086 |
if (!page) { |
94ae8ba71 hugetlb/cgroup: a... |
2087 |
spin_unlock(&hugetlb_lock); |
0c397daea mm, hugetlb: furt... |
2088 |
page = alloc_buddy_huge_page_with_mpol(h, vma, addr); |
8f34af6f9 mm, hugetlb: move... |
2089 2090 |
if (!page) goto out_uncharge_cgroup; |
a88c76954 mm: hugetlb: fix ... |
2091 2092 2093 2094 |
if (!avoid_reserve && vma_has_reserves(vma, gbl_chg)) { SetPagePrivate(page); h->resv_huge_pages--; } |
79dbb2368 hugetlb: move all... |
2095 2096 |
spin_lock(&hugetlb_lock); list_move(&page->lru, &h->hugepage_activelist); |
81a6fcae3 mm, hugetlb: clea... |
2097 |
/* Fall through */ |
68842c9b9 hugetlbfs: fix qu... |
2098 |
} |
81a6fcae3 mm, hugetlb: clea... |
2099 2100 |
hugetlb_cgroup_commit_charge(idx, pages_per_huge_page(h), h_cg, page); spin_unlock(&hugetlb_lock); |
348ea204c hugetlb: split al... |
2101 |
|
90481622d hugepages: fix us... |
2102 |
set_page_private(page, (unsigned long)spool); |
90d8b7e61 hugetlb: enforce ... |
2103 |
|
d85f69b0b mm/hugetlb: alloc... |
2104 2105 |
map_commit = vma_commit_reservation(h, vma, addr); if (unlikely(map_chg > map_commit)) { |
33039678c mm/hugetlb: handl... |
2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 |
/* * The page was added to the reservation map between * vma_needs_reservation and vma_commit_reservation. * This indicates a race with hugetlb_reserve_pages. * Adjust for the subpool count incremented above AND * in hugetlb_reserve_pages for the same page. Also, * the reservation count added in hugetlb_reserve_pages * no longer applies. */ long rsv_adjust; rsv_adjust = hugepage_subpool_put_pages(spool, 1); hugetlb_acct_memory(h, -rsv_adjust); } |
90d8b7e61 hugetlb: enforce ... |
2120 |
return page; |
8f34af6f9 mm, hugetlb: move... |
2121 2122 2123 2124 |
out_uncharge_cgroup: hugetlb_cgroup_uncharge_cgroup(idx, pages_per_huge_page(h), h_cg); out_subpool_put: |
d85f69b0b mm/hugetlb: alloc... |
2125 |
if (map_chg || avoid_reserve) |
8f34af6f9 mm, hugetlb: move... |
2126 |
hugepage_subpool_put_pages(spool, 1); |
feba16e25 mm/hugetlb: add r... |
2127 |
vma_end_reservation(h, vma, addr); |
8f34af6f9 mm, hugetlb: move... |
2128 |
return ERR_PTR(-ENOSPC); |
b45b5bd65 [PATCH] hugepage:... |
2129 |
} |
e24a1307b mm/hugetlb: Allow... |
2130 2131 2132 |
int alloc_bootmem_huge_page(struct hstate *h) __attribute__ ((weak, alias("__alloc_bootmem_huge_page"))); int __alloc_bootmem_huge_page(struct hstate *h) |
aa888a749 hugetlb: support ... |
2133 2134 |
{ struct huge_bootmem_page *m; |
b22610268 mm, hugetlb: fix ... |
2135 |
int nr_nodes, node; |
aa888a749 hugetlb: support ... |
2136 |
|
b22610268 mm, hugetlb: fix ... |
2137 |
for_each_node_mask_to_alloc(h, nr_nodes, node, &node_states[N_MEMORY]) { |
aa888a749 hugetlb: support ... |
2138 |
void *addr; |
eb31d559f memblock: remove ... |
2139 |
addr = memblock_alloc_try_nid_raw( |
8b89a1169 mm/hugetlb.c: use... |
2140 |
huge_page_size(h), huge_page_size(h), |
97ad1087e memblock: replace... |
2141 |
0, MEMBLOCK_ALLOC_ACCESSIBLE, node); |
aa888a749 hugetlb: support ... |
2142 2143 2144 2145 2146 2147 2148 |
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 mm: hugetlb: remo... |
2149 |
goto found; |
aa888a749 hugetlb: support ... |
2150 |
} |
aa888a749 hugetlb: support ... |
2151 2152 2153 2154 |
} return 0; found: |
df994ead5 hugetlb: alloc_bo... |
2155 |
BUG_ON(!IS_ALIGNED(virt_to_phys(m), huge_page_size(h))); |
aa888a749 hugetlb: support ... |
2156 |
/* Put them into a private list first because mem_map is not up yet */ |
330d6e489 mm/hugetlb.c: don... |
2157 |
INIT_LIST_HEAD(&m->list); |
aa888a749 hugetlb: support ... |
2158 2159 2160 2161 |
list_add(&m->list, &huge_boot_pages); m->hstate = h; return 1; } |
d00181b96 mm: use 'unsigned... |
2162 2163 |
static void __init prep_compound_huge_page(struct page *page, unsigned int order) |
18229df5b hugetlb: pull gig... |
2164 2165 2166 2167 2168 2169 |
{ if (unlikely(order > (MAX_ORDER - 1))) prep_compound_gigantic_page(page, order); else prep_compound_page(page, order); } |
aa888a749 hugetlb: support ... |
2170 2171 2172 2173 2174 2175 |
/* 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) { |
40d18ebff mm/hugetlb: remov... |
2176 |
struct page *page = virt_to_page(m); |
aa888a749 hugetlb: support ... |
2177 |
struct hstate *h = m->hstate; |
ee8f248d2 hugetlb: add phys... |
2178 |
|
aa888a749 hugetlb: support ... |
2179 |
WARN_ON(page_count(page) != 1); |
18229df5b hugetlb: pull gig... |
2180 |
prep_compound_huge_page(page, h->order); |
ef5a22be2 mm: hugetlb: init... |
2181 |
WARN_ON(PageReserved(page)); |
aa888a749 hugetlb: support ... |
2182 |
prep_new_huge_page(h, page, page_to_nid(page)); |
af0fb9df7 mm, hugetlb: unif... |
2183 |
put_page(page); /* free it into the hugepage allocator */ |
b0320c7b7 mm: fix negative ... |
2184 2185 2186 2187 2188 2189 |
/* * 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 hugetlb: add hsta... |
2190 |
if (hstate_is_gigantic(h)) |
3dcc0571c mm: correctly upd... |
2191 |
adjust_managed_page_count(page, 1 << h->order); |
520495fe9 mm: hugetlb: yiel... |
2192 |
cond_resched(); |
aa888a749 hugetlb: support ... |
2193 2194 |
} } |
8faa8b077 hugetlb: support ... |
2195 |
static void __init hugetlb_hstate_alloc_pages(struct hstate *h) |
1da177e4c Linux-2.6.12-rc2 |
2196 2197 |
{ unsigned long i; |
f60858f9d hugetlbfs: don't ... |
2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 |
nodemask_t *node_alloc_noretry; if (!hstate_is_gigantic(h)) { /* * Bit mask controlling how hard we retry per-node allocations. * Ignore errors as lower level routines can deal with * node_alloc_noretry == NULL. If this kmalloc fails at boot * time, we are likely in bigger trouble. */ node_alloc_noretry = kmalloc(sizeof(*node_alloc_noretry), GFP_KERNEL); } else { /* allocations done at boot time */ node_alloc_noretry = NULL; } /* bit mask controlling how hard we retry per-node allocations */ if (node_alloc_noretry) nodes_clear(*node_alloc_noretry); |
a55164389 hugetlb: modular ... |
2217 |
|
e5ff21594 hugetlb: multiple... |
2218 |
for (i = 0; i < h->max_huge_pages; ++i) { |
bae7f4ae1 hugetlb: add hsta... |
2219 |
if (hstate_is_gigantic(h)) { |
aa888a749 hugetlb: support ... |
2220 2221 |
if (!alloc_bootmem_huge_page(h)) break; |
0c397daea mm, hugetlb: furt... |
2222 |
} else if (!alloc_pool_huge_page(h, |
f60858f9d hugetlbfs: don't ... |
2223 2224 |
&node_states[N_MEMORY], node_alloc_noretry)) |
1da177e4c Linux-2.6.12-rc2 |
2225 |
break; |
69ed779a1 mm, hugetlb: sche... |
2226 |
cond_resched(); |
1da177e4c Linux-2.6.12-rc2 |
2227 |
} |
d715cf804 mm/hugetlb.c: war... |
2228 2229 |
if (i < h->max_huge_pages) { char buf[32]; |
c6247f72d mm/hugetlb.c: rep... |
2230 |
string_get_size(huge_page_size(h), 1, STRING_UNITS_2, buf, 32); |
d715cf804 mm/hugetlb.c: war... |
2231 2232 2233 2234 2235 |
pr_warn("HugeTLB: allocating %lu of page size %s failed. Only allocated %lu hugepages. ", h->max_huge_pages, buf, i); h->max_huge_pages = i; } |
f60858f9d hugetlbfs: don't ... |
2236 2237 |
kfree(node_alloc_noretry); |
e5ff21594 hugetlb: multiple... |
2238 2239 2240 2241 2242 2243 2244 |
} static void __init hugetlb_init_hstates(void) { struct hstate *h; for_each_hstate(h) { |
641844f56 mm/hugetlb: intro... |
2245 2246 |
if (minimum_order > huge_page_order(h)) minimum_order = huge_page_order(h); |
8faa8b077 hugetlb: support ... |
2247 |
/* oversize hugepages were init'ed in early boot */ |
bae7f4ae1 hugetlb: add hsta... |
2248 |
if (!hstate_is_gigantic(h)) |
8faa8b077 hugetlb: support ... |
2249 |
hugetlb_hstate_alloc_pages(h); |
e5ff21594 hugetlb: multiple... |
2250 |
} |
641844f56 mm/hugetlb: intro... |
2251 |
VM_BUG_ON(minimum_order == UINT_MAX); |
e5ff21594 hugetlb: multiple... |
2252 2253 2254 2255 2256 2257 2258 |
} static void __init report_hugepages(void) { struct hstate *h; for_each_hstate(h) { |
4abd32dba hugetlb: printk c... |
2259 |
char buf[32]; |
c6247f72d mm/hugetlb.c: rep... |
2260 2261 |
string_get_size(huge_page_size(h), 1, STRING_UNITS_2, buf, 32); |
ffb22af5b mm/hugetlb.c: con... |
2262 2263 |
pr_info("HugeTLB registered %s page size, pre-allocated %ld pages ", |
c6247f72d mm/hugetlb.c: rep... |
2264 |
buf, h->free_huge_pages); |
e5ff21594 hugetlb: multiple... |
2265 2266 |
} } |
1da177e4c Linux-2.6.12-rc2 |
2267 |
#ifdef CONFIG_HIGHMEM |
6ae11b278 hugetlb: add node... |
2268 2269 |
static void try_to_free_low(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c Linux-2.6.12-rc2 |
2270 |
{ |
4415cc8df [PATCH] Hugepages... |
2271 |
int i; |
bae7f4ae1 hugetlb: add hsta... |
2272 |
if (hstate_is_gigantic(h)) |
aa888a749 hugetlb: support ... |
2273 |
return; |
6ae11b278 hugetlb: add node... |
2274 |
for_each_node_mask(i, *nodes_allowed) { |
1da177e4c Linux-2.6.12-rc2 |
2275 |
struct page *page, *next; |
a55164389 hugetlb: modular ... |
2276 2277 2278 |
struct list_head *freel = &h->hugepage_freelists[i]; list_for_each_entry_safe(page, next, freel, lru) { if (count >= h->nr_huge_pages) |
6b0c880df hugetlb: fix pool... |
2279 |
return; |
1da177e4c Linux-2.6.12-rc2 |
2280 2281 2282 |
if (PageHighMem(page)) continue; list_del(&page->lru); |
e5ff21594 hugetlb: multiple... |
2283 |
update_and_free_page(h, page); |
a55164389 hugetlb: modular ... |
2284 2285 |
h->free_huge_pages--; h->free_huge_pages_node[page_to_nid(page)]--; |
1da177e4c Linux-2.6.12-rc2 |
2286 2287 2288 2289 |
} } } #else |
6ae11b278 hugetlb: add node... |
2290 2291 |
static inline void try_to_free_low(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c Linux-2.6.12-rc2 |
2292 2293 2294 |
{ } #endif |
20a0307c0 mm: introduce Pag... |
2295 2296 2297 2298 2299 |
/* * 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 hugetlb: add node... |
2300 2301 |
static int adjust_pool_surplus(struct hstate *h, nodemask_t *nodes_allowed, int delta) |
20a0307c0 mm: introduce Pag... |
2302 |
{ |
b22610268 mm, hugetlb: fix ... |
2303 |
int nr_nodes, node; |
20a0307c0 mm: introduce Pag... |
2304 2305 |
VM_BUG_ON(delta != -1 && delta != 1); |
20a0307c0 mm: introduce Pag... |
2306 |
|
b22610268 mm, hugetlb: fix ... |
2307 2308 2309 2310 |
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 hugetlb: balance ... |
2311 |
} |
b22610268 mm, hugetlb: fix ... |
2312 2313 2314 2315 2316 |
} 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 hugetlb: balance ... |
2317 |
} |
b22610268 mm, hugetlb: fix ... |
2318 2319 |
} return 0; |
20a0307c0 mm: introduce Pag... |
2320 |
|
b22610268 mm, hugetlb: fix ... |
2321 2322 2323 2324 |
found: h->surplus_huge_pages += delta; h->surplus_huge_pages_node[node] += delta; return 1; |
20a0307c0 mm: introduce Pag... |
2325 |
} |
a55164389 hugetlb: modular ... |
2326 |
#define persistent_huge_pages(h) (h->nr_huge_pages - h->surplus_huge_pages) |
fd875dca7 hugetlbfs: fix po... |
2327 |
static int set_max_huge_pages(struct hstate *h, unsigned long count, int nid, |
4eb0716e8 hugetlb: allow to... |
2328 |
nodemask_t *nodes_allowed) |
1da177e4c Linux-2.6.12-rc2 |
2329 |
{ |
7893d1d50 hugetlb: Try to g... |
2330 |
unsigned long min_count, ret; |
f60858f9d hugetlbfs: don't ... |
2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 |
NODEMASK_ALLOC(nodemask_t, node_alloc_noretry, GFP_KERNEL); /* * Bit mask controlling how hard we retry per-node allocations. * If we can not allocate the bit mask, do not attempt to allocate * the requested huge pages. */ if (node_alloc_noretry) nodes_clear(*node_alloc_noretry); else return -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
2342 |
|
4eb0716e8 hugetlb: allow to... |
2343 2344 2345 |
spin_lock(&hugetlb_lock); /* |
fd875dca7 hugetlbfs: fix po... |
2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 |
* Check for a node specific request. * Changing node specific huge page count may require a corresponding * change to the global count. In any case, the passed node mask * (nodes_allowed) will restrict alloc/free to the specified node. */ if (nid != NUMA_NO_NODE) { unsigned long old_count = count; count += h->nr_huge_pages - h->nr_huge_pages_node[nid]; /* * User may have specified a large count value which caused the * above calculation to overflow. In this case, they wanted * to allocate as many huge pages as possible. Set count to * largest possible value to align with their intention. */ if (count < old_count) count = ULONG_MAX; } /* |
4eb0716e8 hugetlb: allow to... |
2366 2367 2368 2369 2370 2371 2372 2373 2374 |
* Gigantic pages runtime allocation depend on the capability for large * page range allocation. * If the system does not provide this feature, return an error when * the user tries to allocate gigantic pages but let the user free the * boottime allocated gigantic pages. */ if (hstate_is_gigantic(h) && !IS_ENABLED(CONFIG_CONTIG_ALLOC)) { if (count > persistent_huge_pages(h)) { spin_unlock(&hugetlb_lock); |
f60858f9d hugetlbfs: don't ... |
2375 |
NODEMASK_FREE(node_alloc_noretry); |
4eb0716e8 hugetlb: allow to... |
2376 2377 2378 2379 |
return -EINVAL; } /* Fall through to decrease pool */ } |
aa888a749 hugetlb: support ... |
2380 |
|
7893d1d50 hugetlb: Try to g... |
2381 2382 2383 2384 |
/* * Increase the pool size * First take pages out of surplus state. Then make up the * remaining difference by allocating fresh huge pages. |
d1c3fb1f8 hugetlb: introduc... |
2385 |
* |
0c397daea mm, hugetlb: furt... |
2386 |
* We might race with alloc_surplus_huge_page() here and be unable |
d1c3fb1f8 hugetlb: introduc... |
2387 2388 2389 2390 |
* 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 hugetlb: Try to g... |
2391 |
*/ |
a55164389 hugetlb: modular ... |
2392 |
while (h->surplus_huge_pages && count > persistent_huge_pages(h)) { |
6ae11b278 hugetlb: add node... |
2393 |
if (!adjust_pool_surplus(h, nodes_allowed, -1)) |
7893d1d50 hugetlb: Try to g... |
2394 2395 |
break; } |
a55164389 hugetlb: modular ... |
2396 |
while (count > persistent_huge_pages(h)) { |
7893d1d50 hugetlb: Try to g... |
2397 2398 2399 2400 2401 2402 |
/* * 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); |
649920c6a mm/hugetlb: avoid... |
2403 2404 2405 |
/* yield cpu to avoid soft lockup */ cond_resched(); |
f60858f9d hugetlbfs: don't ... |
2406 2407 |
ret = alloc_pool_huge_page(h, nodes_allowed, node_alloc_noretry); |
7893d1d50 hugetlb: Try to g... |
2408 2409 2410 |
spin_lock(&hugetlb_lock); if (!ret) goto out; |
536240f2b hugetlb: abort a ... |
2411 2412 2413 |
/* Bail for signals. Probably ctrl-c from user */ if (signal_pending(current)) goto out; |
7893d1d50 hugetlb: Try to g... |
2414 |
} |
7893d1d50 hugetlb: Try to g... |
2415 2416 2417 2418 2419 2420 2421 |
/* * 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 hugetlb: introduc... |
2422 2423 2424 2425 |
* * 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 |
0c397daea mm, hugetlb: furt... |
2426 |
* alloc_surplus_huge_page() is checking the global counter, |
d1c3fb1f8 hugetlb: introduc... |
2427 2428 2429 |
* 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 hugetlb: Try to g... |
2430 |
*/ |
a55164389 hugetlb: modular ... |
2431 |
min_count = h->resv_huge_pages + h->nr_huge_pages - h->free_huge_pages; |
6b0c880df hugetlb: fix pool... |
2432 |
min_count = max(count, min_count); |
6ae11b278 hugetlb: add node... |
2433 |
try_to_free_low(h, min_count, nodes_allowed); |
a55164389 hugetlb: modular ... |
2434 |
while (min_count < persistent_huge_pages(h)) { |
6ae11b278 hugetlb: add node... |
2435 |
if (!free_pool_huge_page(h, nodes_allowed, 0)) |
1da177e4c Linux-2.6.12-rc2 |
2436 |
break; |
55f67141a mm: hugetlb: fix ... |
2437 |
cond_resched_lock(&hugetlb_lock); |
1da177e4c Linux-2.6.12-rc2 |
2438 |
} |
a55164389 hugetlb: modular ... |
2439 |
while (count < persistent_huge_pages(h)) { |
6ae11b278 hugetlb: add node... |
2440 |
if (!adjust_pool_surplus(h, nodes_allowed, 1)) |
7893d1d50 hugetlb: Try to g... |
2441 2442 2443 |
break; } out: |
4eb0716e8 hugetlb: allow to... |
2444 |
h->max_huge_pages = persistent_huge_pages(h); |
1da177e4c Linux-2.6.12-rc2 |
2445 |
spin_unlock(&hugetlb_lock); |
4eb0716e8 hugetlb: allow to... |
2446 |
|
f60858f9d hugetlbfs: don't ... |
2447 |
NODEMASK_FREE(node_alloc_noretry); |
4eb0716e8 hugetlb: allow to... |
2448 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
2449 |
} |
a34378701 hugetlb: new sysf... |
2450 2451 2452 2453 2454 2455 2456 2457 2458 |
#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 hugetlb: add per ... |
2459 2460 2461 |
static struct hstate *kobj_to_node_hstate(struct kobject *kobj, int *nidp); static struct hstate *kobj_to_hstate(struct kobject *kobj, int *nidp) |
a34378701 hugetlb: new sysf... |
2462 2463 |
{ int i; |
9a3052306 hugetlb: add per ... |
2464 |
|
a34378701 hugetlb: new sysf... |
2465 |
for (i = 0; i < HUGE_MAX_HSTATE; i++) |
9a3052306 hugetlb: add per ... |
2466 2467 2468 |
if (hstate_kobjs[i] == kobj) { if (nidp) *nidp = NUMA_NO_NODE; |
a34378701 hugetlb: new sysf... |
2469 |
return &hstates[i]; |
9a3052306 hugetlb: add per ... |
2470 2471 2472 |
} return kobj_to_node_hstate(kobj, nidp); |
a34378701 hugetlb: new sysf... |
2473 |
} |
06808b082 hugetlb: derive h... |
2474 |
static ssize_t nr_hugepages_show_common(struct kobject *kobj, |
a34378701 hugetlb: new sysf... |
2475 2476 |
struct kobj_attribute *attr, char *buf) { |
9a3052306 hugetlb: add per ... |
2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 |
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 hugetlb: new sysf... |
2489 |
} |
adbe8726d hugetlb: do not a... |
2490 |
|
238d3c13f mm, hugetlb: gene... |
2491 2492 2493 |
static ssize_t __nr_hugepages_store_common(bool obey_mempolicy, struct hstate *h, int nid, unsigned long count, size_t len) |
a34378701 hugetlb: new sysf... |
2494 2495 |
{ int err; |
2d0adf7e0 mm/hugetlb: get r... |
2496 |
nodemask_t nodes_allowed, *n_mask; |
a34378701 hugetlb: new sysf... |
2497 |
|
2d0adf7e0 mm/hugetlb: get r... |
2498 2499 |
if (hstate_is_gigantic(h) && !gigantic_page_runtime_supported()) return -EINVAL; |
adbe8726d hugetlb: do not a... |
2500 |
|
9a3052306 hugetlb: add per ... |
2501 2502 2503 2504 2505 |
if (nid == NUMA_NO_NODE) { /* * global hstate attribute */ if (!(obey_mempolicy && |
2d0adf7e0 mm/hugetlb: get r... |
2506 2507 2508 2509 2510 |
init_nodemask_of_mempolicy(&nodes_allowed))) n_mask = &node_states[N_MEMORY]; else n_mask = &nodes_allowed; } else { |
9a3052306 hugetlb: add per ... |
2511 |
/* |
fd875dca7 hugetlbfs: fix po... |
2512 2513 |
* Node specific request. count adjustment happens in * set_max_huge_pages() after acquiring hugetlb_lock. |
9a3052306 hugetlb: add per ... |
2514 |
*/ |
2d0adf7e0 mm/hugetlb: get r... |
2515 2516 |
init_nodemask_of_node(&nodes_allowed, nid); n_mask = &nodes_allowed; |
fd875dca7 hugetlbfs: fix po... |
2517 |
} |
9a3052306 hugetlb: add per ... |
2518 |
|
2d0adf7e0 mm/hugetlb: get r... |
2519 |
err = set_max_huge_pages(h, count, nid, n_mask); |
06808b082 hugetlb: derive h... |
2520 |
|
4eb0716e8 hugetlb: allow to... |
2521 |
return err ? err : len; |
06808b082 hugetlb: derive h... |
2522 |
} |
238d3c13f mm, hugetlb: gene... |
2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 |
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 hugetlb: derive h... |
2539 2540 2541 2542 2543 2544 2545 2546 2547 |
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 mm, hugetlb: gene... |
2548 |
return nr_hugepages_store_common(false, kobj, buf, len); |
a34378701 hugetlb: new sysf... |
2549 2550 |
} HSTATE_ATTR(nr_hugepages); |
06808b082 hugetlb: derive h... |
2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 |
#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 mm, hugetlb: gene... |
2566 |
return nr_hugepages_store_common(true, kobj, buf, len); |
06808b082 hugetlb: derive h... |
2567 2568 2569 |
} HSTATE_ATTR(nr_hugepages_mempolicy); #endif |
a34378701 hugetlb: new sysf... |
2570 2571 2572 |
static ssize_t nr_overcommit_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
9a3052306 hugetlb: add per ... |
2573 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701 hugetlb: new sysf... |
2574 2575 2576 |
return sprintf(buf, "%lu ", h->nr_overcommit_huge_pages); } |
adbe8726d hugetlb: do not a... |
2577 |
|
a34378701 hugetlb: new sysf... |
2578 2579 2580 2581 2582 |
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 hugetlb: add per ... |
2583 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701 hugetlb: new sysf... |
2584 |
|
bae7f4ae1 hugetlb: add hsta... |
2585 |
if (hstate_is_gigantic(h)) |
adbe8726d hugetlb: do not a... |
2586 |
return -EINVAL; |
3dbb95f78 mm: replace stric... |
2587 |
err = kstrtoul(buf, 10, &input); |
a34378701 hugetlb: new sysf... |
2588 |
if (err) |
73ae31e59 hugetlb: fix hand... |
2589 |
return err; |
a34378701 hugetlb: new sysf... |
2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 |
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 hugetlb: add per ... |
2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 |
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 hugetlb: new sysf... |
2614 2615 2616 2617 2618 2619 |
} HSTATE_ATTR_RO(free_hugepages); static ssize_t resv_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
9a3052306 hugetlb: add per ... |
2620 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701 hugetlb: new sysf... |
2621 2622 2623 2624 2625 2626 2627 2628 |
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 hugetlb: add per ... |
2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 |
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 hugetlb: new sysf... |
2641 2642 2643 2644 2645 2646 2647 2648 2649 |
} 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 hugetlb: derive h... |
2650 2651 2652 |
#ifdef CONFIG_NUMA &nr_hugepages_mempolicy_attr.attr, #endif |
a34378701 hugetlb: new sysf... |
2653 2654 |
NULL, }; |
67e5ed969 mm/hugetlb.c: con... |
2655 |
static const struct attribute_group hstate_attr_group = { |
a34378701 hugetlb: new sysf... |
2656 2657 |
.attrs = hstate_attrs, }; |
094e9539b hugetlb: fix sect... |
2658 2659 |
static int hugetlb_sysfs_add_hstate(struct hstate *h, struct kobject *parent, struct kobject **hstate_kobjs, |
67e5ed969 mm/hugetlb.c: con... |
2660 |
const struct attribute_group *hstate_attr_group) |
a34378701 hugetlb: new sysf... |
2661 2662 |
{ int retval; |
972dc4de1 hugetlb: add an i... |
2663 |
int hi = hstate_index(h); |
a34378701 hugetlb: new sysf... |
2664 |
|
9a3052306 hugetlb: add per ... |
2665 2666 |
hstate_kobjs[hi] = kobject_create_and_add(h->name, parent); if (!hstate_kobjs[hi]) |
a34378701 hugetlb: new sysf... |
2667 |
return -ENOMEM; |
9a3052306 hugetlb: add per ... |
2668 |
retval = sysfs_create_group(hstate_kobjs[hi], hstate_attr_group); |
a34378701 hugetlb: new sysf... |
2669 |
if (retval) |
9a3052306 hugetlb: add per ... |
2670 |
kobject_put(hstate_kobjs[hi]); |
a34378701 hugetlb: new sysf... |
2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 |
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 hugetlb: add per ... |
2685 2686 |
err = hugetlb_sysfs_add_hstate(h, hugepages_kobj, hstate_kobjs, &hstate_attr_group); |
a34378701 hugetlb: new sysf... |
2687 |
if (err) |
ffb22af5b mm/hugetlb.c: con... |
2688 |
pr_err("Hugetlb: Unable to add hstate %s", h->name); |
a34378701 hugetlb: new sysf... |
2689 2690 |
} } |
9a3052306 hugetlb: add per ... |
2691 2692 2693 2694 |
#ifdef CONFIG_NUMA /* * node_hstate/s - associate per node hstate attributes, via their kobjects, |
10fbcf4c6 convert 'memory' ... |
2695 2696 2697 |
* 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 hugetlb: add per ... |
2698 2699 2700 2701 2702 2703 |
* the base kernel, on the hugetlb module. */ struct node_hstate { struct kobject *hugepages_kobj; struct kobject *hstate_kobjs[HUGE_MAX_HSTATE]; }; |
b4e289a6a mm/hugetlb: make ... |
2704 |
static struct node_hstate node_hstates[MAX_NUMNODES]; |
9a3052306 hugetlb: add per ... |
2705 2706 |
/* |
10fbcf4c6 convert 'memory' ... |
2707 |
* A subset of global hstate attributes for node devices |
9a3052306 hugetlb: add per ... |
2708 2709 2710 2711 2712 2713 2714 |
*/ static struct attribute *per_node_hstate_attrs[] = { &nr_hugepages_attr.attr, &free_hugepages_attr.attr, &surplus_hugepages_attr.attr, NULL, }; |
67e5ed969 mm/hugetlb.c: con... |
2715 |
static const struct attribute_group per_node_hstate_attr_group = { |
9a3052306 hugetlb: add per ... |
2716 2717 2718 2719 |
.attrs = per_node_hstate_attrs, }; /* |
10fbcf4c6 convert 'memory' ... |
2720 |
* kobj_to_node_hstate - lookup global hstate for node device hstate attr kobj. |
9a3052306 hugetlb: add per ... |
2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 |
* 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 convert 'memory' ... |
2743 |
* Unregister hstate attributes from a single node device. |
9a3052306 hugetlb: add per ... |
2744 2745 |
* No-op if no hstate attributes attached. */ |
3cd8b44fa hugetlb: fix spar... |
2746 |
static void hugetlb_unregister_node(struct node *node) |
9a3052306 hugetlb: add per ... |
2747 2748 |
{ struct hstate *h; |
10fbcf4c6 convert 'memory' ... |
2749 |
struct node_hstate *nhs = &node_hstates[node->dev.id]; |
9a3052306 hugetlb: add per ... |
2750 2751 |
if (!nhs->hugepages_kobj) |
9b5e5d0fd hugetlb: use only... |
2752 |
return; /* no hstate attributes */ |
9a3052306 hugetlb: add per ... |
2753 |
|
972dc4de1 hugetlb: add an i... |
2754 2755 2756 2757 2758 |
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 hugetlb: add per ... |
2759 |
} |
972dc4de1 hugetlb: add an i... |
2760 |
} |
9a3052306 hugetlb: add per ... |
2761 2762 2763 2764 |
kobject_put(nhs->hugepages_kobj); nhs->hugepages_kobj = NULL; } |
9a3052306 hugetlb: add per ... |
2765 2766 |
/* |
10fbcf4c6 convert 'memory' ... |
2767 |
* Register hstate attributes for a single node device. |
9a3052306 hugetlb: add per ... |
2768 2769 |
* No-op if attributes already registered. */ |
3cd8b44fa hugetlb: fix spar... |
2770 |
static void hugetlb_register_node(struct node *node) |
9a3052306 hugetlb: add per ... |
2771 2772 |
{ struct hstate *h; |
10fbcf4c6 convert 'memory' ... |
2773 |
struct node_hstate *nhs = &node_hstates[node->dev.id]; |
9a3052306 hugetlb: add per ... |
2774 2775 2776 2777 2778 2779 |
int err; if (nhs->hugepages_kobj) return; /* already allocated */ nhs->hugepages_kobj = kobject_create_and_add("hugepages", |
10fbcf4c6 convert 'memory' ... |
2780 |
&node->dev.kobj); |
9a3052306 hugetlb: add per ... |
2781 2782 2783 2784 2785 2786 2787 2788 |
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 mm/hugetlb.c: con... |
2789 2790 2791 |
pr_err("Hugetlb: Unable to add hstate %s for node %d ", h->name, node->dev.id); |
9a3052306 hugetlb: add per ... |
2792 2793 2794 2795 2796 2797 2798 |
hugetlb_unregister_node(node); break; } } } /* |
9b5e5d0fd hugetlb: use only... |
2799 |
* hugetlb init time: register hstate attributes for all registered node |
10fbcf4c6 convert 'memory' ... |
2800 2801 |
* devices of nodes that have memory. All on-line nodes should have * registered their associated device by this time. |
9a3052306 hugetlb: add per ... |
2802 |
*/ |
7d9ca0004 hugetlb: hugetlb_... |
2803 |
static void __init hugetlb_register_all_nodes(void) |
9a3052306 hugetlb: add per ... |
2804 2805 |
{ int nid; |
8cebfcd07 hugetlb: use N_ME... |
2806 |
for_each_node_state(nid, N_MEMORY) { |
8732794b1 numa: convert sta... |
2807 |
struct node *node = node_devices[nid]; |
10fbcf4c6 convert 'memory' ... |
2808 |
if (node->dev.id == nid) |
9a3052306 hugetlb: add per ... |
2809 2810 2811 2812 |
hugetlb_register_node(node); } /* |
10fbcf4c6 convert 'memory' ... |
2813 |
* Let the node device driver know we're here so it can |
9a3052306 hugetlb: add per ... |
2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 |
* [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; } |
9a3052306 hugetlb: add per ... |
2828 2829 2830 |
static void hugetlb_register_all_nodes(void) { } #endif |
a34378701 hugetlb: new sysf... |
2831 2832 |
static int __init hugetlb_init(void) { |
8382d914e mm, hugetlb: impr... |
2833 |
int i; |
457c1b27e hugetlb: ensure h... |
2834 |
if (!hugepages_supported()) |
0ef89d25d mm/hugetlb: don't... |
2835 |
return 0; |
a34378701 hugetlb: new sysf... |
2836 |
|
e11bfbfcb hugetlb: override... |
2837 |
if (!size_to_hstate(default_hstate_size)) { |
d715cf804 mm/hugetlb.c: war... |
2838 2839 2840 2841 2842 |
if (default_hstate_size != 0) { pr_err("HugeTLB: unsupported default_hugepagesz %lu. Reverting to %lu ", default_hstate_size, HPAGE_SIZE); } |
e11bfbfcb hugetlb: override... |
2843 2844 2845 |
default_hstate_size = HPAGE_SIZE; if (!size_to_hstate(default_hstate_size)) hugetlb_add_hstate(HUGETLB_PAGE_ORDER); |
a34378701 hugetlb: new sysf... |
2846 |
} |
972dc4de1 hugetlb: add an i... |
2847 |
default_hstate_idx = hstate_index(size_to_hstate(default_hstate_size)); |
f8b74815a mm/hugetlb.c: fix... |
2848 2849 2850 2851 |
if (default_hstate_max_huge_pages) { if (!default_hstate.max_huge_pages) default_hstate.max_huge_pages = default_hstate_max_huge_pages; } |
a34378701 hugetlb: new sysf... |
2852 2853 |
hugetlb_init_hstates(); |
aa888a749 hugetlb: support ... |
2854 |
gather_bootmem_prealloc(); |
a34378701 hugetlb: new sysf... |
2855 2856 2857 |
report_hugepages(); hugetlb_sysfs_init(); |
9a3052306 hugetlb: add per ... |
2858 |
hugetlb_register_all_nodes(); |
7179e7bf4 mm/hugetlb: creat... |
2859 |
hugetlb_cgroup_file_init(); |
9a3052306 hugetlb: add per ... |
2860 |
|
8382d914e mm, hugetlb: impr... |
2861 2862 2863 2864 2865 |
#ifdef CONFIG_SMP num_fault_mutexes = roundup_pow_of_two(8 * num_possible_cpus()); #else num_fault_mutexes = 1; #endif |
c672c7f29 mm/hugetlb: expos... |
2866 |
hugetlb_fault_mutex_table = |
6da2ec560 treewide: kmalloc... |
2867 2868 |
kmalloc_array(num_fault_mutexes, sizeof(struct mutex), GFP_KERNEL); |
c672c7f29 mm/hugetlb: expos... |
2869 |
BUG_ON(!hugetlb_fault_mutex_table); |
8382d914e mm, hugetlb: impr... |
2870 2871 |
for (i = 0; i < num_fault_mutexes; i++) |
c672c7f29 mm/hugetlb: expos... |
2872 |
mutex_init(&hugetlb_fault_mutex_table[i]); |
a34378701 hugetlb: new sysf... |
2873 2874 |
return 0; } |
3e89e1c5e hugetlb: make mm ... |
2875 |
subsys_initcall(hugetlb_init); |
a34378701 hugetlb: new sysf... |
2876 2877 |
/* Should be called on processing a hugepagesz=... option */ |
9fee021d1 mm/hugetlb: intro... |
2878 2879 2880 2881 |
void __init hugetlb_bad_size(void) { parsed_valid_hugepagesz = false; } |
d00181b96 mm: use 'unsigned... |
2882 |
void __init hugetlb_add_hstate(unsigned int order) |
a34378701 hugetlb: new sysf... |
2883 2884 |
{ struct hstate *h; |
8faa8b077 hugetlb: support ... |
2885 |
unsigned long i; |
a34378701 hugetlb: new sysf... |
2886 |
if (size_to_hstate(PAGE_SIZE << order)) { |
598d80914 mm: convert pr_wa... |
2887 2888 |
pr_warn("hugepagesz= specified twice, ignoring "); |
a34378701 hugetlb: new sysf... |
2889 2890 |
return; } |
47d38344a hugetlb: rename m... |
2891 |
BUG_ON(hugetlb_max_hstate >= HUGE_MAX_HSTATE); |
a34378701 hugetlb: new sysf... |
2892 |
BUG_ON(order == 0); |
47d38344a hugetlb: rename m... |
2893 |
h = &hstates[hugetlb_max_hstate++]; |
a34378701 hugetlb: new sysf... |
2894 2895 |
h->order = order; h->mask = ~((1ULL << (order + PAGE_SHIFT)) - 1); |
8faa8b077 hugetlb: support ... |
2896 2897 2898 2899 |
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 hugetlb: add a li... |
2900 |
INIT_LIST_HEAD(&h->hugepage_activelist); |
54f18d352 mm/hugetlb.c: use... |
2901 2902 |
h->next_nid_to_alloc = first_memory_node; h->next_nid_to_free = first_memory_node; |
a34378701 hugetlb: new sysf... |
2903 2904 |
snprintf(h->name, HSTATE_NAME_LEN, "hugepages-%lukB", huge_page_size(h)/1024); |
8faa8b077 hugetlb: support ... |
2905 |
|
a34378701 hugetlb: new sysf... |
2906 2907 |
parsed_hstate = h; } |
e11bfbfcb hugetlb: override... |
2908 |
static int __init hugetlb_nrpages_setup(char *s) |
a34378701 hugetlb: new sysf... |
2909 2910 |
{ unsigned long *mhp; |
8faa8b077 hugetlb: support ... |
2911 |
static unsigned long *last_mhp; |
a34378701 hugetlb: new sysf... |
2912 |
|
9fee021d1 mm/hugetlb: intro... |
2913 2914 2915 2916 2917 2918 2919 |
if (!parsed_valid_hugepagesz) { pr_warn("hugepages = %s preceded by " "an unsupported hugepagesz, ignoring ", s); parsed_valid_hugepagesz = true; return 1; } |
a34378701 hugetlb: new sysf... |
2920 |
/* |
47d38344a hugetlb: rename m... |
2921 |
* !hugetlb_max_hstate means we haven't parsed a hugepagesz= parameter yet, |
a34378701 hugetlb: new sysf... |
2922 2923 |
* so this hugepages= parameter goes to the "default hstate". */ |
9fee021d1 mm/hugetlb: intro... |
2924 |
else if (!hugetlb_max_hstate) |
a34378701 hugetlb: new sysf... |
2925 2926 2927 |
mhp = &default_hstate_max_huge_pages; else mhp = &parsed_hstate->max_huge_pages; |
8faa8b077 hugetlb: support ... |
2928 |
if (mhp == last_mhp) { |
598d80914 mm: convert pr_wa... |
2929 2930 |
pr_warn("hugepages= specified twice without interleaving hugepagesz=, ignoring "); |
8faa8b077 hugetlb: support ... |
2931 2932 |
return 1; } |
a34378701 hugetlb: new sysf... |
2933 2934 |
if (sscanf(s, "%lu", mhp) <= 0) *mhp = 0; |
8faa8b077 hugetlb: support ... |
2935 2936 2937 2938 2939 |
/* * 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 hugetlb: rename m... |
2940 |
if (hugetlb_max_hstate && parsed_hstate->order >= MAX_ORDER) |
8faa8b077 hugetlb: support ... |
2941 2942 2943 |
hugetlb_hstate_alloc_pages(parsed_hstate); last_mhp = mhp; |
a34378701 hugetlb: new sysf... |
2944 2945 |
return 1; } |
e11bfbfcb hugetlb: override... |
2946 2947 2948 2949 2950 2951 2952 2953 |
__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 hugetlb: new sysf... |
2954 |
|
8a2134605 hugetlb: fix CONF... |
2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 |
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 |
af7786b20 mm/hugetlb: fix a... |
2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 |
static int proc_hugetlb_doulongvec_minmax(struct ctl_table *table, int write, void *buffer, size_t *length, loff_t *ppos, unsigned long *out) { struct ctl_table dup_table; /* * In order to avoid races with __do_proc_doulongvec_minmax(), we * can duplicate the @table and alter the duplicate of it. */ dup_table = *table; dup_table.data = out; return proc_doulongvec_minmax(&dup_table, write, buffer, length, ppos); } |
06808b082 hugetlb: derive h... |
2982 2983 2984 |
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 Linux-2.6.12-rc2 |
2985 |
{ |
e5ff21594 hugetlb: multiple... |
2986 |
struct hstate *h = &default_hstate; |
238d3c13f mm, hugetlb: gene... |
2987 |
unsigned long tmp = h->max_huge_pages; |
08d4a2465 hugetlb: check th... |
2988 |
int ret; |
e5ff21594 hugetlb: multiple... |
2989 |
|
457c1b27e hugetlb: ensure h... |
2990 |
if (!hugepages_supported()) |
86613628b mm/hugetlb: use E... |
2991 |
return -EOPNOTSUPP; |
457c1b27e hugetlb: ensure h... |
2992 |
|
af7786b20 mm/hugetlb: fix a... |
2993 2994 |
ret = proc_hugetlb_doulongvec_minmax(table, write, buffer, length, ppos, &tmp); |
08d4a2465 hugetlb: check th... |
2995 2996 |
if (ret) goto out; |
e5ff21594 hugetlb: multiple... |
2997 |
|
238d3c13f mm, hugetlb: gene... |
2998 2999 3000 |
if (write) ret = __nr_hugepages_store_common(obey_mempolicy, h, NUMA_NO_NODE, tmp, *length); |
08d4a2465 hugetlb: check th... |
3001 3002 |
out: return ret; |
1da177e4c Linux-2.6.12-rc2 |
3003 |
} |
396faf030 Allow huge page a... |
3004 |
|
06808b082 hugetlb: derive h... |
3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 |
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 hugetlb: add lock... |
3021 |
int hugetlb_overcommit_handler(struct ctl_table *table, int write, |
8d65af789 sysctl: remove "s... |
3022 |
void __user *buffer, |
a3d0c6aa1 hugetlb: add lock... |
3023 3024 |
size_t *length, loff_t *ppos) { |
a55164389 hugetlb: modular ... |
3025 |
struct hstate *h = &default_hstate; |
e5ff21594 hugetlb: multiple... |
3026 |
unsigned long tmp; |
08d4a2465 hugetlb: check th... |
3027 |
int ret; |
e5ff21594 hugetlb: multiple... |
3028 |
|
457c1b27e hugetlb: ensure h... |
3029 |
if (!hugepages_supported()) |
86613628b mm/hugetlb: use E... |
3030 |
return -EOPNOTSUPP; |
457c1b27e hugetlb: ensure h... |
3031 |
|
c033a93c0 hugetlbfs: correc... |
3032 |
tmp = h->nr_overcommit_huge_pages; |
e5ff21594 hugetlb: multiple... |
3033 |
|
bae7f4ae1 hugetlb: add hsta... |
3034 |
if (write && hstate_is_gigantic(h)) |
adbe8726d hugetlb: do not a... |
3035 |
return -EINVAL; |
af7786b20 mm/hugetlb: fix a... |
3036 3037 |
ret = proc_hugetlb_doulongvec_minmax(table, write, buffer, length, ppos, &tmp); |
08d4a2465 hugetlb: check th... |
3038 3039 |
if (ret) goto out; |
e5ff21594 hugetlb: multiple... |
3040 3041 3042 3043 3044 3045 |
if (write) { spin_lock(&hugetlb_lock); h->nr_overcommit_huge_pages = tmp; spin_unlock(&hugetlb_lock); } |
08d4a2465 hugetlb: check th... |
3046 3047 |
out: return ret; |
a3d0c6aa1 hugetlb: add lock... |
3048 |
} |
1da177e4c Linux-2.6.12-rc2 |
3049 |
#endif /* CONFIG_SYSCTL */ |
e1759c215 proc: switch /pro... |
3050 |
void hugetlb_report_meminfo(struct seq_file *m) |
1da177e4c Linux-2.6.12-rc2 |
3051 |
{ |
fcb2b0c57 mm: show total hu... |
3052 3053 |
struct hstate *h; unsigned long total = 0; |
457c1b27e hugetlb: ensure h... |
3054 3055 |
if (!hugepages_supported()) return; |
fcb2b0c57 mm: show total hu... |
3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 |
for_each_hstate(h) { unsigned long count = h->nr_huge_pages; total += (PAGE_SIZE << huge_page_order(h)) * count; if (h == &default_hstate) seq_printf(m, "HugePages_Total: %5lu " "HugePages_Free: %5lu " "HugePages_Rsvd: %5lu " "HugePages_Surp: %5lu " "Hugepagesize: %8lu kB ", count, h->free_huge_pages, h->resv_huge_pages, h->surplus_huge_pages, (PAGE_SIZE << huge_page_order(h)) / 1024); } seq_printf(m, "Hugetlb: %8lu kB ", total / 1024); |
1da177e4c Linux-2.6.12-rc2 |
3083 3084 3085 3086 |
} int hugetlb_report_node_meminfo(int nid, char *buf) { |
a55164389 hugetlb: modular ... |
3087 |
struct hstate *h = &default_hstate; |
457c1b27e hugetlb: ensure h... |
3088 3089 |
if (!hugepages_supported()) return 0; |
1da177e4c Linux-2.6.12-rc2 |
3090 3091 3092 |
return sprintf(buf, "Node %d HugePages_Total: %5u " |
a1de09195 hugetlb: indicate... |
3093 3094 3095 3096 |
"Node %d HugePages_Free: %5u " "Node %d HugePages_Surp: %5u ", |
a55164389 hugetlb: modular ... |
3097 3098 3099 |
nid, h->nr_huge_pages_node[nid], nid, h->free_huge_pages_node[nid], nid, h->surplus_huge_pages_node[nid]); |
1da177e4c Linux-2.6.12-rc2 |
3100 |
} |
949f7ec57 mm, hugetlb: incl... |
3101 3102 3103 3104 |
void hugetlb_show_meminfo(void) { struct hstate *h; int nid; |
457c1b27e hugetlb: ensure h... |
3105 3106 |
if (!hugepages_supported()) return; |
949f7ec57 mm, hugetlb: incl... |
3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 |
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)); } |
5d317b2b6 mm: hugetlb: proc... |
3117 3118 3119 3120 3121 3122 |
void hugetlb_report_usage(struct seq_file *m, struct mm_struct *mm) { seq_printf(m, "HugetlbPages:\t%8lu kB ", atomic_long_read(&mm->hugetlb_usage) << (PAGE_SHIFT - 10)); } |
1da177e4c Linux-2.6.12-rc2 |
3123 3124 3125 |
/* Return the number pages of memory we physically have, in PAGE_SIZE units. */ unsigned long hugetlb_total_pages(void) { |
d00285884 mm/hugetlb: fix t... |
3126 3127 3128 3129 3130 3131 |
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 Linux-2.6.12-rc2 |
3132 |
} |
1da177e4c Linux-2.6.12-rc2 |
3133 |
|
a55164389 hugetlb: modular ... |
3134 |
static int hugetlb_acct_memory(struct hstate *h, long delta) |
fc1b8a73d hugetlb: move hug... |
3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 |
{ 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 hugetlb: modular ... |
3157 |
if (gather_surplus_pages(h, delta) < 0) |
fc1b8a73d hugetlb: move hug... |
3158 |
goto out; |
a55164389 hugetlb: modular ... |
3159 3160 |
if (delta > cpuset_mems_nr(h->free_huge_pages_node)) { return_unused_surplus_pages(h, delta); |
fc1b8a73d hugetlb: move hug... |
3161 3162 3163 3164 3165 3166 |
goto out; } } ret = 0; if (delta < 0) |
a55164389 hugetlb: modular ... |
3167 |
return_unused_surplus_pages(h, (unsigned long) -delta); |
fc1b8a73d hugetlb: move hug... |
3168 3169 3170 3171 3172 |
out: spin_unlock(&hugetlb_lock); return ret; } |
84afd99b8 hugetlb reservati... |
3173 3174 |
static void hugetlb_vm_op_open(struct vm_area_struct *vma) { |
f522c3ac0 mm, hugetlb: chan... |
3175 |
struct resv_map *resv = vma_resv_map(vma); |
84afd99b8 hugetlb reservati... |
3176 3177 3178 3179 3180 |
/* * 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 Fix common misspe... |
3181 |
* has a reference to the reservation map it cannot disappear until |
84afd99b8 hugetlb reservati... |
3182 3183 3184 |
* after this open call completes. It is therefore safe to take a * new reference here without additional locking. */ |
4e35f4838 mm, hugetlb: use ... |
3185 |
if (resv && is_vma_resv_set(vma, HPAGE_RESV_OWNER)) |
f522c3ac0 mm, hugetlb: chan... |
3186 |
kref_get(&resv->refs); |
84afd99b8 hugetlb reservati... |
3187 |
} |
a1e78772d hugetlb: reserve ... |
3188 3189 |
static void hugetlb_vm_op_close(struct vm_area_struct *vma) { |
a55164389 hugetlb: modular ... |
3190 |
struct hstate *h = hstate_vma(vma); |
f522c3ac0 mm, hugetlb: chan... |
3191 |
struct resv_map *resv = vma_resv_map(vma); |
90481622d hugepages: fix us... |
3192 |
struct hugepage_subpool *spool = subpool_vma(vma); |
4e35f4838 mm, hugetlb: use ... |
3193 |
unsigned long reserve, start, end; |
1c5ecae3a hugetlbfs: add mi... |
3194 |
long gbl_reserve; |
84afd99b8 hugetlb reservati... |
3195 |
|
4e35f4838 mm, hugetlb: use ... |
3196 3197 |
if (!resv || !is_vma_resv_set(vma, HPAGE_RESV_OWNER)) return; |
84afd99b8 hugetlb reservati... |
3198 |
|
4e35f4838 mm, hugetlb: use ... |
3199 3200 |
start = vma_hugecache_offset(h, vma, vma->vm_start); end = vma_hugecache_offset(h, vma, vma->vm_end); |
84afd99b8 hugetlb reservati... |
3201 |
|
4e35f4838 mm, hugetlb: use ... |
3202 |
reserve = (end - start) - region_count(resv, start, end); |
84afd99b8 hugetlb reservati... |
3203 |
|
4e35f4838 mm, hugetlb: use ... |
3204 3205 3206 |
kref_put(&resv->refs, resv_map_release); if (reserve) { |
1c5ecae3a hugetlbfs: add mi... |
3207 3208 3209 3210 3211 3212 |
/* * Decrement reserve counts. The global reserve count may be * adjusted if the subpool has a minimum size. */ gbl_reserve = hugepage_subpool_put_pages(spool, reserve); hugetlb_acct_memory(h, -gbl_reserve); |
84afd99b8 hugetlb reservati... |
3213 |
} |
a1e78772d hugetlb: reserve ... |
3214 |
} |
31383c686 mm, hugetlbfs: in... |
3215 3216 3217 3218 3219 3220 |
static int hugetlb_vm_op_split(struct vm_area_struct *vma, unsigned long addr) { if (addr & ~(huge_page_mask(hstate_vma(vma)))) return -EINVAL; return 0; } |
05ea88608 mm, hugetlbfs: in... |
3221 3222 3223 3224 3225 3226 |
static unsigned long hugetlb_vm_op_pagesize(struct vm_area_struct *vma) { struct hstate *hstate = hstate_vma(vma); return 1UL << huge_page_shift(hstate); } |
1da177e4c Linux-2.6.12-rc2 |
3227 3228 3229 3230 3231 3232 |
/* * 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. */ |
b3ec9f33a mm: change return... |
3233 |
static vm_fault_t hugetlb_vm_op_fault(struct vm_fault *vmf) |
1da177e4c Linux-2.6.12-rc2 |
3234 3235 |
{ BUG(); |
d0217ac04 mm: fault feedbac... |
3236 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
3237 |
} |
eec3636ad ipc/shm.c add ->p... |
3238 3239 3240 3241 3242 3243 3244 |
/* * When a new function is introduced to vm_operations_struct and added * to hugetlb_vm_ops, please consider adding the function to shm_vm_ops. * This is because under System V memory model, mappings created via * shmget/shmat with "huge page" specified are backed by hugetlbfs files, * their original vm_ops are overwritten with shm_vm_ops. */ |
f0f37e2f7 const: mark struc... |
3245 |
const struct vm_operations_struct hugetlb_vm_ops = { |
d0217ac04 mm: fault feedbac... |
3246 |
.fault = hugetlb_vm_op_fault, |
84afd99b8 hugetlb reservati... |
3247 |
.open = hugetlb_vm_op_open, |
a1e78772d hugetlb: reserve ... |
3248 |
.close = hugetlb_vm_op_close, |
31383c686 mm, hugetlbfs: in... |
3249 |
.split = hugetlb_vm_op_split, |
05ea88608 mm, hugetlbfs: in... |
3250 |
.pagesize = hugetlb_vm_op_pagesize, |
1da177e4c Linux-2.6.12-rc2 |
3251 |
}; |
1e8f889b1 [PATCH] Hugetlb: ... |
3252 3253 |
static pte_t make_huge_pte(struct vm_area_struct *vma, struct page *page, int writable) |
63551ae0f [PATCH] Hugepage ... |
3254 3255 |
{ pte_t entry; |
1e8f889b1 [PATCH] Hugetlb: ... |
3256 |
if (writable) { |
106c992a5 mm/hugetlb: add m... |
3257 3258 |
entry = huge_pte_mkwrite(huge_pte_mkdirty(mk_huge_pte(page, vma->vm_page_prot))); |
63551ae0f [PATCH] Hugepage ... |
3259 |
} else { |
106c992a5 mm/hugetlb: add m... |
3260 3261 |
entry = huge_pte_wrprotect(mk_huge_pte(page, vma->vm_page_prot)); |
63551ae0f [PATCH] Hugepage ... |
3262 3263 3264 |
} entry = pte_mkyoung(entry); entry = pte_mkhuge(entry); |
d9ed9faac mm: add new arch_... |
3265 |
entry = arch_make_huge_pte(entry, vma, page, writable); |
63551ae0f [PATCH] Hugepage ... |
3266 3267 3268 |
return entry; } |
1e8f889b1 [PATCH] Hugetlb: ... |
3269 3270 3271 3272 |
static void set_huge_ptep_writable(struct vm_area_struct *vma, unsigned long address, pte_t *ptep) { pte_t entry; |
106c992a5 mm/hugetlb: add m... |
3273 |
entry = huge_pte_mkwrite(huge_pte_mkdirty(huge_ptep_get(ptep))); |
32f84528f mm: hugetlb: fix ... |
3274 |
if (huge_ptep_set_access_flags(vma, address, ptep, entry, 1)) |
4b3073e1c MM: Pass a PTE po... |
3275 |
update_mmu_cache(vma, address, ptep); |
1e8f889b1 [PATCH] Hugetlb: ... |
3276 |
} |
d5ed7444d mm/hugetlb: expor... |
3277 |
bool is_hugetlb_entry_migration(pte_t pte) |
4a705fef9 hugetlb: fix copy... |
3278 3279 3280 3281 |
{ swp_entry_t swp; if (huge_pte_none(pte) || pte_present(pte)) |
d5ed7444d mm/hugetlb: expor... |
3282 |
return false; |
4a705fef9 hugetlb: fix copy... |
3283 3284 |
swp = pte_to_swp_entry(pte); if (non_swap_entry(swp) && is_migration_entry(swp)) |
d5ed7444d mm/hugetlb: expor... |
3285 |
return true; |
4a705fef9 hugetlb: fix copy... |
3286 |
else |
d5ed7444d mm/hugetlb: expor... |
3287 |
return false; |
4a705fef9 hugetlb: fix copy... |
3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 |
} 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 [PATCH] Hugetlb: ... |
3302 |
|
63551ae0f [PATCH] Hugepage ... |
3303 3304 3305 |
int copy_hugetlb_page_range(struct mm_struct *dst, struct mm_struct *src, struct vm_area_struct *vma) { |
5e41540c8 hugetlbfs: fix ke... |
3306 |
pte_t *src_pte, *dst_pte, entry, dst_entry; |
63551ae0f [PATCH] Hugepage ... |
3307 |
struct page *ptepage; |
1c59827d1 [PATCH] mm: huget... |
3308 |
unsigned long addr; |
1e8f889b1 [PATCH] Hugetlb: ... |
3309 |
int cow; |
a55164389 hugetlb: modular ... |
3310 3311 |
struct hstate *h = hstate_vma(vma); unsigned long sz = huge_page_size(h); |
ac46d4f3c mm/mmu_notifier: ... |
3312 |
struct mmu_notifier_range range; |
e8569dd29 mm/hugetlb.c: cal... |
3313 |
int ret = 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
3314 3315 |
cow = (vma->vm_flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE; |
63551ae0f [PATCH] Hugepage ... |
3316 |
|
ac46d4f3c mm/mmu_notifier: ... |
3317 |
if (cow) { |
7269f9999 mm/mmu_notifier: ... |
3318 |
mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, vma, src, |
6f4f13e8d mm/mmu_notifier: ... |
3319 |
vma->vm_start, |
ac46d4f3c mm/mmu_notifier: ... |
3320 3321 3322 |
vma->vm_end); mmu_notifier_invalidate_range_start(&range); } |
e8569dd29 mm/hugetlb.c: cal... |
3323 |
|
a55164389 hugetlb: modular ... |
3324 |
for (addr = vma->vm_start; addr < vma->vm_end; addr += sz) { |
cb900f412 mm, hugetlb: conv... |
3325 |
spinlock_t *src_ptl, *dst_ptl; |
7868a2087 mm/hugetlb: add s... |
3326 |
src_pte = huge_pte_offset(src, addr, sz); |
c74df32c7 [PATCH] mm: ptd_a... |
3327 3328 |
if (!src_pte) continue; |
a55164389 hugetlb: modular ... |
3329 |
dst_pte = huge_pte_alloc(dst, addr, sz); |
e8569dd29 mm/hugetlb.c: cal... |
3330 3331 3332 3333 |
if (!dst_pte) { ret = -ENOMEM; break; } |
c5c99429f fix hugepages lea... |
3334 |
|
5e41540c8 hugetlbfs: fix ke... |
3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 |
/* * If the pagetables are shared don't copy or take references. * dst_pte == src_pte is the common case of src/dest sharing. * * However, src could have 'unshared' and dst shares with * another vma. If dst_pte !none, this implies sharing. * Check here before taking page table lock, and once again * after taking the lock below. */ dst_entry = huge_ptep_get(dst_pte); if ((dst_pte == src_pte) || !huge_pte_none(dst_entry)) |
c5c99429f fix hugepages lea... |
3346 |
continue; |
cb900f412 mm, hugetlb: conv... |
3347 3348 3349 |
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 hugetlb: fix copy... |
3350 |
entry = huge_ptep_get(src_pte); |
5e41540c8 hugetlbfs: fix ke... |
3351 3352 3353 3354 3355 3356 3357 |
dst_entry = huge_ptep_get(dst_pte); if (huge_pte_none(entry) || !huge_pte_none(dst_entry)) { /* * Skip if src entry none. Also, skip in the * unlikely case dst entry !none as this implies * sharing with another vma. */ |
4a705fef9 hugetlb: fix copy... |
3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 |
; } 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); |
e5251fd43 mm/hugetlb: intro... |
3370 3371 |
set_huge_swap_pte_at(src, addr, src_pte, entry, sz); |
4a705fef9 hugetlb: fix copy... |
3372 |
} |
e5251fd43 mm/hugetlb: intro... |
3373 |
set_huge_swap_pte_at(dst, addr, dst_pte, entry, sz); |
4a705fef9 hugetlb: fix copy... |
3374 |
} else { |
34ee645e8 mmu_notifier: cal... |
3375 |
if (cow) { |
0f10851ea mm/mmu_notifier: ... |
3376 3377 3378 3379 3380 |
/* * No need to notify as we are downgrading page * table protection not changing it to point * to a new page. * |
ad56b738c docs/vm: rename d... |
3381 |
* See Documentation/vm/mmu_notifier.rst |
0f10851ea mm/mmu_notifier: ... |
3382 |
*/ |
7f2e9525b hugetlbfs: common... |
3383 |
huge_ptep_set_wrprotect(src, addr, src_pte); |
34ee645e8 mmu_notifier: cal... |
3384 |
} |
0253d634e mm: hugetlb: fix ... |
3385 |
entry = huge_ptep_get(src_pte); |
1c59827d1 [PATCH] mm: huget... |
3386 3387 |
ptepage = pte_page(entry); get_page(ptepage); |
53f9263ba mm: rework mapcou... |
3388 |
page_dup_rmap(ptepage, true); |
1c59827d1 [PATCH] mm: huget... |
3389 |
set_huge_pte_at(dst, addr, dst_pte, entry); |
5d317b2b6 mm: hugetlb: proc... |
3390 |
hugetlb_count_add(pages_per_huge_page(h), dst); |
1c59827d1 [PATCH] mm: huget... |
3391 |
} |
cb900f412 mm, hugetlb: conv... |
3392 3393 |
spin_unlock(src_ptl); spin_unlock(dst_ptl); |
63551ae0f [PATCH] Hugepage ... |
3394 |
} |
63551ae0f [PATCH] Hugepage ... |
3395 |
|
e8569dd29 mm/hugetlb.c: cal... |
3396 |
if (cow) |
ac46d4f3c mm/mmu_notifier: ... |
3397 |
mmu_notifier_invalidate_range_end(&range); |
e8569dd29 mm/hugetlb.c: cal... |
3398 3399 |
return ret; |
63551ae0f [PATCH] Hugepage ... |
3400 |
} |
24669e584 hugetlb: use mmu_... |
3401 3402 3403 |
void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma, unsigned long start, unsigned long end, struct page *ref_page) |
63551ae0f [PATCH] Hugepage ... |
3404 3405 3406 |
{ struct mm_struct *mm = vma->vm_mm; unsigned long address; |
c7546f8f0 [PATCH] Fix hugep... |
3407 |
pte_t *ptep; |
63551ae0f [PATCH] Hugepage ... |
3408 |
pte_t pte; |
cb900f412 mm, hugetlb: conv... |
3409 |
spinlock_t *ptl; |
63551ae0f [PATCH] Hugepage ... |
3410 |
struct page *page; |
a55164389 hugetlb: modular ... |
3411 3412 |
struct hstate *h = hstate_vma(vma); unsigned long sz = huge_page_size(h); |
ac46d4f3c mm/mmu_notifier: ... |
3413 |
struct mmu_notifier_range range; |
a55164389 hugetlb: modular ... |
3414 |
|
63551ae0f [PATCH] Hugepage ... |
3415 |
WARN_ON(!is_vm_hugetlb_page(vma)); |
a55164389 hugetlb: modular ... |
3416 3417 |
BUG_ON(start & ~huge_page_mask(h)); BUG_ON(end & ~huge_page_mask(h)); |
63551ae0f [PATCH] Hugepage ... |
3418 |
|
07e326610 mm: add tlb_remov... |
3419 3420 3421 3422 |
/* * This is a hugetlb vma, all the pte entries should point * to huge page. */ |
ed6a79352 asm-generic/tlb, ... |
3423 |
tlb_change_page_size(tlb, sz); |
24669e584 hugetlb: use mmu_... |
3424 |
tlb_start_vma(tlb, vma); |
dff11abe2 hugetlb: take PMD... |
3425 3426 3427 3428 |
/* * If sharing possible, alert mmu notifiers of worst case. */ |
6f4f13e8d mm/mmu_notifier: ... |
3429 3430 |
mmu_notifier_range_init(&range, MMU_NOTIFY_UNMAP, 0, vma, mm, start, end); |
ac46d4f3c mm/mmu_notifier: ... |
3431 3432 |
adjust_range_if_pmd_sharing_possible(vma, &range.start, &range.end); mmu_notifier_invalidate_range_start(&range); |
569f48b85 mm: hugetlb: fix ... |
3433 |
address = start; |
569f48b85 mm: hugetlb: fix ... |
3434 |
for (; address < end; address += sz) { |
7868a2087 mm/hugetlb: add s... |
3435 |
ptep = huge_pte_offset(mm, address, sz); |
4c8872659 [PATCH] hugetlb: ... |
3436 |
if (!ptep) |
c7546f8f0 [PATCH] Fix hugep... |
3437 |
continue; |
cb900f412 mm, hugetlb: conv... |
3438 |
ptl = huge_pte_lock(h, mm, ptep); |
31d49da5a mm/hugetlb: simpl... |
3439 3440 |
if (huge_pmd_unshare(mm, &address, ptep)) { spin_unlock(ptl); |
dff11abe2 hugetlb: take PMD... |
3441 3442 3443 3444 |
/* * We just unmapped a page of PMDs by clearing a PUD. * The caller's TLB flush range should cover this area. */ |
31d49da5a mm/hugetlb: simpl... |
3445 3446 |
continue; } |
39dde65c9 [PATCH] shared pa... |
3447 |
|
6629326b8 mm: hugetlb: clea... |
3448 |
pte = huge_ptep_get(ptep); |
31d49da5a mm/hugetlb: simpl... |
3449 3450 3451 3452 |
if (huge_pte_none(pte)) { spin_unlock(ptl); continue; } |
6629326b8 mm: hugetlb: clea... |
3453 3454 |
/* |
9fbc1f635 mm/hugetlb: add m... |
3455 3456 |
* Migrating hugepage or HWPoisoned hugepage is already * unmapped and its refcount is dropped, so just clear pte here. |
6629326b8 mm: hugetlb: clea... |
3457 |
*/ |
9fbc1f635 mm/hugetlb: add m... |
3458 |
if (unlikely(!pte_present(pte))) { |
9386fac34 mm/hugetlb: allow... |
3459 |
huge_pte_clear(mm, address, ptep, sz); |
31d49da5a mm/hugetlb: simpl... |
3460 3461 |
spin_unlock(ptl); continue; |
8c4894c6b hwpoison, hugetlb... |
3462 |
} |
6629326b8 mm: hugetlb: clea... |
3463 3464 |
page = pte_page(pte); |
04f2cbe35 hugetlb: guarante... |
3465 3466 3467 3468 3469 3470 |
/* * 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) { |
31d49da5a mm/hugetlb: simpl... |
3471 3472 3473 3474 |
if (page != ref_page) { spin_unlock(ptl); continue; } |
04f2cbe35 hugetlb: guarante... |
3475 3476 3477 3478 3479 3480 3481 |
/* * 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 [PATCH] Fix hugep... |
3482 |
pte = huge_ptep_get_and_clear(mm, address, ptep); |
b528e4b64 mm/hugetlb: add t... |
3483 |
tlb_remove_huge_tlb_entry(h, tlb, ptep, address); |
106c992a5 mm/hugetlb: add m... |
3484 |
if (huge_pte_dirty(pte)) |
6649a3863 [PATCH] hugetlb: ... |
3485 |
set_page_dirty(page); |
9e81130b7 mm: hugetlb: bail... |
3486 |
|
5d317b2b6 mm: hugetlb: proc... |
3487 |
hugetlb_count_sub(pages_per_huge_page(h), mm); |
d281ee614 rmap: add argumen... |
3488 |
page_remove_rmap(page, true); |
31d49da5a mm/hugetlb: simpl... |
3489 |
|
cb900f412 mm, hugetlb: conv... |
3490 |
spin_unlock(ptl); |
e77b0852b mm/mmu_gather: tr... |
3491 |
tlb_remove_page_size(tlb, page, huge_page_size(h)); |
31d49da5a mm/hugetlb: simpl... |
3492 3493 3494 3495 3496 |
/* * Bail out after unmapping reference page if supplied */ if (ref_page) break; |
fe1668ae5 [PATCH] enforce p... |
3497 |
} |
ac46d4f3c mm/mmu_notifier: ... |
3498 |
mmu_notifier_invalidate_range_end(&range); |
24669e584 hugetlb: use mmu_... |
3499 |
tlb_end_vma(tlb, vma); |
1da177e4c Linux-2.6.12-rc2 |
3500 |
} |
63551ae0f [PATCH] Hugepage ... |
3501 |
|
d833352a4 mm: hugetlbfs: cl... |
3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 |
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 mm: convert i_mma... |
3514 |
* is to clear it before releasing the i_mmap_rwsem. This works |
d833352a4 mm: hugetlbfs: cl... |
3515 |
* because in the context this is called, the VMA is about to be |
c8c06efa8 mm: convert i_mma... |
3516 |
* destroyed and the i_mmap_rwsem is held. |
d833352a4 mm: hugetlbfs: cl... |
3517 3518 3519 |
*/ vma->vm_flags &= ~VM_MAYSHARE; } |
502717f4e [PATCH] hugetlb: ... |
3520 |
void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start, |
04f2cbe35 hugetlb: guarante... |
3521 |
unsigned long end, struct page *ref_page) |
502717f4e [PATCH] hugetlb: ... |
3522 |
{ |
24669e584 hugetlb: use mmu_... |
3523 3524 |
struct mm_struct *mm; struct mmu_gather tlb; |
dff11abe2 hugetlb: take PMD... |
3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 |
unsigned long tlb_start = start; unsigned long tlb_end = end; /* * If shared PMDs were possibly used within this vma range, adjust * start/end for worst case tlb flushing. * Note that we can not be sure if PMDs are shared until we try to * unmap pages. However, we want to make sure TLB flushing covers * the largest possible range. */ adjust_range_if_pmd_sharing_possible(vma, &tlb_start, &tlb_end); |
24669e584 hugetlb: use mmu_... |
3536 3537 |
mm = vma->vm_mm; |
dff11abe2 hugetlb: take PMD... |
3538 |
tlb_gather_mmu(&tlb, mm, tlb_start, tlb_end); |
24669e584 hugetlb: use mmu_... |
3539 |
__unmap_hugepage_range(&tlb, vma, start, end, ref_page); |
dff11abe2 hugetlb: take PMD... |
3540 |
tlb_finish_mmu(&tlb, tlb_start, tlb_end); |
502717f4e [PATCH] hugetlb: ... |
3541 |
} |
04f2cbe35 hugetlb: guarante... |
3542 3543 3544 3545 3546 3547 |
/* * 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 mm,hugetlb: make ... |
3548 3549 |
static void unmap_ref_private(struct mm_struct *mm, struct vm_area_struct *vma, struct page *page, unsigned long address) |
04f2cbe35 hugetlb: guarante... |
3550 |
{ |
7526674de hugetlb: make unm... |
3551 |
struct hstate *h = hstate_vma(vma); |
04f2cbe35 hugetlb: guarante... |
3552 3553 |
struct vm_area_struct *iter_vma; struct address_space *mapping; |
04f2cbe35 hugetlb: guarante... |
3554 3555 3556 3557 3558 3559 |
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 hugetlb: make unm... |
3560 |
address = address & huge_page_mask(h); |
36e4f20af hugetlb: do not u... |
3561 3562 |
pgoff = ((address - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff; |
93c76a3d4 file_inode(f)->i_... |
3563 |
mapping = vma->vm_file->f_mapping; |
04f2cbe35 hugetlb: guarante... |
3564 |
|
4eb2b1dcd hugetlb: acquire ... |
3565 3566 3567 3568 3569 |
/* * 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 mm: use new helpe... |
3570 |
i_mmap_lock_write(mapping); |
6b2dbba8b mm: replace vma p... |
3571 |
vma_interval_tree_foreach(iter_vma, &mapping->i_mmap, pgoff, pgoff) { |
04f2cbe35 hugetlb: guarante... |
3572 3573 3574 3575 3576 |
/* Do not unmap the current VMA */ if (iter_vma == vma) continue; /* |
2f84a8990 mm: hugetlbfs: sk... |
3577 3578 3579 3580 3581 3582 3583 3584 |
* Shared VMAs have their own reserves and do not affect * MAP_PRIVATE accounting but it is possible that a shared * VMA is using the same page so check and skip such VMAs. */ if (iter_vma->vm_flags & VM_MAYSHARE) continue; /* |
04f2cbe35 hugetlb: guarante... |
3585 3586 3587 3588 3589 3590 3591 |
* 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 hugetlb: use mmu_... |
3592 3593 |
unmap_hugepage_range(iter_vma, address, address + huge_page_size(h), page); |
04f2cbe35 hugetlb: guarante... |
3594 |
} |
83cde9e8b mm: use new helpe... |
3595 |
i_mmap_unlock_write(mapping); |
04f2cbe35 hugetlb: guarante... |
3596 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
3597 3598 |
/* * Hugetlb_cow() should be called with page lock of the original hugepage held. |
ef009b25f hugetlb: clarify ... |
3599 3600 3601 |
* 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 hugetlb, rmap: ad... |
3602 |
*/ |
2b7403035 mm: Change return... |
3603 |
static vm_fault_t hugetlb_cow(struct mm_struct *mm, struct vm_area_struct *vma, |
974e6d66b mm, hugetlbfs: pa... |
3604 |
unsigned long address, pte_t *ptep, |
3999f52e3 mm/hugetlb.c: use... |
3605 |
struct page *pagecache_page, spinlock_t *ptl) |
1e8f889b1 [PATCH] Hugetlb: ... |
3606 |
{ |
3999f52e3 mm/hugetlb.c: use... |
3607 |
pte_t pte; |
a55164389 hugetlb: modular ... |
3608 |
struct hstate *h = hstate_vma(vma); |
1e8f889b1 [PATCH] Hugetlb: ... |
3609 |
struct page *old_page, *new_page; |
2b7403035 mm: Change return... |
3610 3611 |
int outside_reserve = 0; vm_fault_t ret = 0; |
974e6d66b mm, hugetlbfs: pa... |
3612 |
unsigned long haddr = address & huge_page_mask(h); |
ac46d4f3c mm/mmu_notifier: ... |
3613 |
struct mmu_notifier_range range; |
1e8f889b1 [PATCH] Hugetlb: ... |
3614 |
|
3999f52e3 mm/hugetlb.c: use... |
3615 |
pte = huge_ptep_get(ptep); |
1e8f889b1 [PATCH] Hugetlb: ... |
3616 |
old_page = pte_page(pte); |
04f2cbe35 hugetlb: guarante... |
3617 |
retry_avoidcopy: |
1e8f889b1 [PATCH] Hugetlb: ... |
3618 3619 |
/* If no-one else is actually using this page, avoid the copy * and just make the page writable */ |
37a2140dc mm, hugetlb: do n... |
3620 |
if (page_mapcount(old_page) == 1 && PageAnon(old_page)) { |
5a49973d7 mm: thp: refix fa... |
3621 |
page_move_anon_rmap(old_page, vma); |
5b7a1d406 mm, hugetlbfs: re... |
3622 |
set_huge_ptep_writable(vma, haddr, ptep); |
83c54070e mm: fault feedbac... |
3623 |
return 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
3624 |
} |
04f2cbe35 hugetlb: guarante... |
3625 3626 3627 3628 3629 3630 3631 3632 3633 |
/* * 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 mm, hugetlb: remo... |
3634 |
if (is_vma_resv_set(vma, HPAGE_RESV_OWNER) && |
04f2cbe35 hugetlb: guarante... |
3635 3636 |
old_page != pagecache_page) outside_reserve = 1; |
09cbfeaf1 mm, fs: get rid o... |
3637 |
get_page(old_page); |
b76c8cfbf hugetlb: prevent ... |
3638 |
|
ad4404a22 mm,hugetlb: simpl... |
3639 3640 3641 3642 |
/* * Drop page table lock as buddy allocator may be called. It will * be acquired again before returning to the caller, as expected. */ |
cb900f412 mm, hugetlb: conv... |
3643 |
spin_unlock(ptl); |
5b7a1d406 mm, hugetlbfs: re... |
3644 |
new_page = alloc_huge_page(vma, haddr, outside_reserve); |
1e8f889b1 [PATCH] Hugetlb: ... |
3645 |
|
2fc39cec6 hugetlb: debit qu... |
3646 |
if (IS_ERR(new_page)) { |
04f2cbe35 hugetlb: guarante... |
3647 3648 3649 3650 3651 3652 3653 3654 |
/* * 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) { |
09cbfeaf1 mm, fs: get rid o... |
3655 |
put_page(old_page); |
04f2cbe35 hugetlb: guarante... |
3656 |
BUG_ON(huge_pte_none(pte)); |
5b7a1d406 mm, hugetlbfs: re... |
3657 |
unmap_ref_private(mm, vma, old_page, haddr); |
2f4612af4 mm,hugetlb: make ... |
3658 3659 |
BUG_ON(huge_pte_none(pte)); spin_lock(ptl); |
5b7a1d406 mm, hugetlbfs: re... |
3660 |
ptep = huge_pte_offset(mm, haddr, huge_page_size(h)); |
2f4612af4 mm,hugetlb: make ... |
3661 3662 3663 3664 3665 3666 3667 3668 |
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 hugetlb: guarante... |
3669 |
} |
2b7403035 mm: Change return... |
3670 |
ret = vmf_error(PTR_ERR(new_page)); |
ad4404a22 mm,hugetlb: simpl... |
3671 |
goto out_release_old; |
1e8f889b1 [PATCH] Hugetlb: ... |
3672 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
3673 3674 3675 3676 |
/* * When the original hugepage is shared one, it does not have * anon_vma prepared. */ |
44e2aa937 mm/hugetlb.c: add... |
3677 |
if (unlikely(anon_vma_prepare(vma))) { |
ad4404a22 mm,hugetlb: simpl... |
3678 3679 |
ret = VM_FAULT_OOM; goto out_release_all; |
44e2aa937 mm/hugetlb.c: add... |
3680 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
3681 |
|
974e6d66b mm, hugetlbfs: pa... |
3682 |
copy_user_huge_page(new_page, old_page, address, vma, |
47ad8475c thp: clear_copy_h... |
3683 |
pages_per_huge_page(h)); |
0ed361dec mm: fix PageUptod... |
3684 |
__SetPageUptodate(new_page); |
1e8f889b1 [PATCH] Hugetlb: ... |
3685 |
|
7269f9999 mm/mmu_notifier: ... |
3686 |
mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, vma, mm, haddr, |
6f4f13e8d mm/mmu_notifier: ... |
3687 |
haddr + huge_page_size(h)); |
ac46d4f3c mm/mmu_notifier: ... |
3688 |
mmu_notifier_invalidate_range_start(&range); |
ad4404a22 mm,hugetlb: simpl... |
3689 |
|
b76c8cfbf hugetlb: prevent ... |
3690 |
/* |
cb900f412 mm, hugetlb: conv... |
3691 |
* Retake the page table lock to check for racing updates |
b76c8cfbf hugetlb: prevent ... |
3692 3693 |
* before the page tables are altered */ |
cb900f412 mm, hugetlb: conv... |
3694 |
spin_lock(ptl); |
5b7a1d406 mm, hugetlbfs: re... |
3695 |
ptep = huge_pte_offset(mm, haddr, huge_page_size(h)); |
a9af0c5df mm/hugetlb.c: add... |
3696 |
if (likely(ptep && pte_same(huge_ptep_get(ptep), pte))) { |
07443a85a mm, hugetlb: retu... |
3697 |
ClearPagePrivate(new_page); |
1e8f889b1 [PATCH] Hugetlb: ... |
3698 |
/* Break COW */ |
5b7a1d406 mm, hugetlbfs: re... |
3699 |
huge_ptep_clear_flush(vma, haddr, ptep); |
ac46d4f3c mm/mmu_notifier: ... |
3700 |
mmu_notifier_invalidate_range(mm, range.start, range.end); |
5b7a1d406 mm, hugetlbfs: re... |
3701 |
set_huge_pte_at(mm, haddr, ptep, |
1e8f889b1 [PATCH] Hugetlb: ... |
3702 |
make_huge_pte(vma, new_page, 1)); |
d281ee614 rmap: add argumen... |
3703 |
page_remove_rmap(old_page, true); |
5b7a1d406 mm, hugetlbfs: re... |
3704 |
hugepage_add_new_anon_rmap(new_page, vma, haddr); |
cb6acd01e hugetlbfs: fix ra... |
3705 |
set_page_huge_active(new_page); |
1e8f889b1 [PATCH] Hugetlb: ... |
3706 3707 3708 |
/* Make the old page be freed below */ new_page = old_page; } |
cb900f412 mm, hugetlb: conv... |
3709 |
spin_unlock(ptl); |
ac46d4f3c mm/mmu_notifier: ... |
3710 |
mmu_notifier_invalidate_range_end(&range); |
ad4404a22 mm,hugetlb: simpl... |
3711 |
out_release_all: |
5b7a1d406 mm, hugetlbfs: re... |
3712 |
restore_reserve_on_error(h, vma, haddr, new_page); |
09cbfeaf1 mm, fs: get rid o... |
3713 |
put_page(new_page); |
ad4404a22 mm,hugetlb: simpl... |
3714 |
out_release_old: |
09cbfeaf1 mm, fs: get rid o... |
3715 |
put_page(old_page); |
8312034f3 mm, hugetlb: grab... |
3716 |
|
ad4404a22 mm,hugetlb: simpl... |
3717 3718 |
spin_lock(ptl); /* Caller expects lock to be held */ return ret; |
1e8f889b1 [PATCH] Hugetlb: ... |
3719 |
} |
04f2cbe35 hugetlb: guarante... |
3720 |
/* Return the pagecache page at a given address within a VMA */ |
a55164389 hugetlb: modular ... |
3721 3722 |
static struct page *hugetlbfs_pagecache_page(struct hstate *h, struct vm_area_struct *vma, unsigned long address) |
04f2cbe35 hugetlb: guarante... |
3723 3724 |
{ struct address_space *mapping; |
e7c4b0bfd huge page private... |
3725 |
pgoff_t idx; |
04f2cbe35 hugetlb: guarante... |
3726 3727 |
mapping = vma->vm_file->f_mapping; |
a55164389 hugetlb: modular ... |
3728 |
idx = vma_hugecache_offset(h, vma, address); |
04f2cbe35 hugetlb: guarante... |
3729 3730 3731 |
return find_lock_page(mapping, idx); } |
3ae77f43b mm: hugetlbfs_pag... |
3732 3733 3734 3735 3736 |
/* * 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 mm: follow_hugetl... |
3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 |
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; } |
ab76ad540 hugetlbfs: New hu... |
3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 |
int huge_add_to_page_cache(struct page *page, struct address_space *mapping, pgoff_t idx) { struct inode *inode = mapping->host; struct hstate *h = hstate_inode(inode); int err = add_to_page_cache(page, mapping, idx, GFP_KERNEL); if (err) return err; ClearPagePrivate(page); |
22146c3ce hugetlbfs: dirty ... |
3761 3762 3763 3764 3765 |
/* * set page dirty so that it will not be removed from cache/file * by non-hugetlbfs specific code paths. */ set_page_dirty(page); |
ab76ad540 hugetlbfs: New hu... |
3766 3767 3768 3769 3770 |
spin_lock(&inode->i_lock); inode->i_blocks += blocks_per_huge_page(h); spin_unlock(&inode->i_lock); return 0; } |
2b7403035 mm: Change return... |
3771 3772 3773 3774 |
static vm_fault_t hugetlb_no_page(struct mm_struct *mm, struct vm_area_struct *vma, struct address_space *mapping, pgoff_t idx, unsigned long address, pte_t *ptep, unsigned int flags) |
ac9b9c667 [PATCH] Fix handl... |
3775 |
{ |
a55164389 hugetlb: modular ... |
3776 |
struct hstate *h = hstate_vma(vma); |
2b7403035 mm: Change return... |
3777 |
vm_fault_t ret = VM_FAULT_SIGBUS; |
409eb8c26 mm/hugetlb.c: und... |
3778 |
int anon_rmap = 0; |
4c8872659 [PATCH] hugetlb: ... |
3779 |
unsigned long size; |
4c8872659 [PATCH] hugetlb: ... |
3780 |
struct page *page; |
1e8f889b1 [PATCH] Hugetlb: ... |
3781 |
pte_t new_pte; |
cb900f412 mm, hugetlb: conv... |
3782 |
spinlock_t *ptl; |
285b8dcaa mm, hugetlbfs: pa... |
3783 |
unsigned long haddr = address & huge_page_mask(h); |
cb6acd01e hugetlbfs: fix ra... |
3784 |
bool new_page = false; |
4c8872659 [PATCH] hugetlb: ... |
3785 |
|
04f2cbe35 hugetlb: guarante... |
3786 3787 3788 |
/* * 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 Fix common misspe... |
3789 |
* COW. Warn that such a situation has occurred as it may not be obvious |
04f2cbe35 hugetlb: guarante... |
3790 3791 |
*/ if (is_vma_resv_set(vma, HPAGE_RESV_UNMAPPED)) { |
910154d52 mm/hugetlb: huget... |
3792 3793 |
pr_warn_ratelimited("PID %d killed due to inadequate hugepage pool ", |
ffb22af5b mm/hugetlb.c: con... |
3794 |
current->pid); |
04f2cbe35 hugetlb: guarante... |
3795 3796 |
return ret; } |
4c8872659 [PATCH] hugetlb: ... |
3797 |
/* |
e7c580977 hugetlbfs: revert... |
3798 3799 |
* Use page lock to guard against racing truncation * before we get page_table_lock. |
4c8872659 [PATCH] hugetlb: ... |
3800 |
*/ |
6bda666a0 [PATCH] hugepages... |
3801 3802 3803 |
retry: page = find_lock_page(mapping, idx); if (!page) { |
e7c580977 hugetlbfs: revert... |
3804 3805 3806 |
size = i_size_read(mapping->host) >> huge_page_shift(h); if (idx >= size) goto out; |
1a1aad8a9 userfaultfd: huge... |
3807 3808 3809 3810 3811 3812 3813 |
/* * Check for page in userfault range */ if (userfaultfd_missing(vma)) { u32 hash; struct vm_fault vmf = { .vma = vma, |
285b8dcaa mm, hugetlbfs: pa... |
3814 |
.address = haddr, |
1a1aad8a9 userfaultfd: huge... |
3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 |
.flags = flags, /* * Hard to debug if it ends up being * used by a callee that assumes * something about the other * uninitialized fields... same as in * memory.c */ }; /* |
ddeaab32a hugetlbfs: revert... |
3826 3827 3828 |
* hugetlb_fault_mutex must be dropped before * handling userfault. Reacquire after handling * fault to make calling code simpler. |
1a1aad8a9 userfaultfd: huge... |
3829 |
*/ |
1b426bac6 hugetlb: use same... |
3830 |
hash = hugetlb_fault_mutex_hash(h, mapping, idx, haddr); |
1a1aad8a9 userfaultfd: huge... |
3831 3832 3833 3834 3835 |
mutex_unlock(&hugetlb_fault_mutex_table[hash]); ret = handle_userfault(&vmf, VM_UFFD_MISSING); mutex_lock(&hugetlb_fault_mutex_table[hash]); goto out; } |
285b8dcaa mm, hugetlbfs: pa... |
3836 |
page = alloc_huge_page(vma, haddr, 0); |
2fc39cec6 hugetlb: debit qu... |
3837 |
if (IS_ERR(page)) { |
4643d67e8 hugetlbfs: fix hu... |
3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 |
/* * Returning error will result in faulting task being * sent SIGBUS. The hugetlb fault mutex prevents two * tasks from racing to fault in the same page which * could result in false unable to allocate errors. * Page migration does not take the fault mutex, but * does a clear then write of pte's under page table * lock. Page fault code could race with migration, * notice the clear pte and try to allocate a page * here. Before returning error, get ptl and make * sure there really is no pte entry. */ ptl = huge_pte_lock(h, mm, ptep); if (!huge_pte_none(huge_ptep_get(ptep))) { ret = 0; spin_unlock(ptl); goto out; } spin_unlock(ptl); |
2b7403035 mm: Change return... |
3857 |
ret = vmf_error(PTR_ERR(page)); |
6bda666a0 [PATCH] hugepages... |
3858 3859 |
goto out; } |
47ad8475c thp: clear_copy_h... |
3860 |
clear_huge_page(page, address, pages_per_huge_page(h)); |
0ed361dec mm: fix PageUptod... |
3861 |
__SetPageUptodate(page); |
cb6acd01e hugetlbfs: fix ra... |
3862 |
new_page = true; |
ac9b9c667 [PATCH] Fix handl... |
3863 |
|
f83a275db mm: account for M... |
3864 |
if (vma->vm_flags & VM_MAYSHARE) { |
ab76ad540 hugetlbfs: New hu... |
3865 |
int err = huge_add_to_page_cache(page, mapping, idx); |
6bda666a0 [PATCH] hugepages... |
3866 3867 |
if (err) { put_page(page); |
6bda666a0 [PATCH] hugepages... |
3868 3869 3870 3871 |
if (err == -EEXIST) goto retry; goto out; } |
23be7468e hugetlb: fix infi... |
3872 |
} else { |
6bda666a0 [PATCH] hugepages... |
3873 |
lock_page(page); |
0fe6e20b9 hugetlb, rmap: ad... |
3874 3875 3876 3877 |
if (unlikely(anon_vma_prepare(vma))) { ret = VM_FAULT_OOM; goto backout_unlocked; } |
409eb8c26 mm/hugetlb.c: und... |
3878 |
anon_rmap = 1; |
23be7468e hugetlb: fix infi... |
3879 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
3880 |
} else { |
998b4382c hugetlb: fix meta... |
3881 3882 3883 3884 3885 3886 |
/* * 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 mm: hugetlb: fix ... |
3887 |
ret = VM_FAULT_HWPOISON | |
972dc4de1 hugetlb: add an i... |
3888 |
VM_FAULT_SET_HINDEX(hstate_index(h)); |
998b4382c hugetlb: fix meta... |
3889 3890 |
goto backout_unlocked; } |
6bda666a0 [PATCH] hugepages... |
3891 |
} |
1e8f889b1 [PATCH] Hugetlb: ... |
3892 |
|
57303d801 hugetlbfs: alloca... |
3893 3894 3895 3896 3897 3898 |
/* * 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. */ |
5e9113731 mm/hugetlb: add c... |
3899 |
if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) { |
285b8dcaa mm, hugetlbfs: pa... |
3900 |
if (vma_needs_reservation(h, vma, haddr) < 0) { |
2b26736c8 allocate structur... |
3901 3902 3903 |
ret = VM_FAULT_OOM; goto backout_unlocked; } |
5e9113731 mm/hugetlb: add c... |
3904 |
/* Just decrements count, does not deallocate */ |
285b8dcaa mm, hugetlbfs: pa... |
3905 |
vma_end_reservation(h, vma, haddr); |
5e9113731 mm/hugetlb: add c... |
3906 |
} |
57303d801 hugetlbfs: alloca... |
3907 |
|
8bea80520 mm/hugetlb.c: use... |
3908 |
ptl = huge_pte_lock(h, mm, ptep); |
e7c580977 hugetlbfs: revert... |
3909 3910 3911 |
size = i_size_read(mapping->host) >> huge_page_shift(h); if (idx >= size) goto backout; |
4c8872659 [PATCH] hugetlb: ... |
3912 |
|
83c54070e mm: fault feedbac... |
3913 |
ret = 0; |
7f2e9525b hugetlbfs: common... |
3914 |
if (!huge_pte_none(huge_ptep_get(ptep))) |
4c8872659 [PATCH] hugetlb: ... |
3915 |
goto backout; |
07443a85a mm, hugetlb: retu... |
3916 3917 |
if (anon_rmap) { ClearPagePrivate(page); |
285b8dcaa mm, hugetlbfs: pa... |
3918 |
hugepage_add_new_anon_rmap(page, vma, haddr); |
ac7149045 mm: fix 'ERROR: d... |
3919 |
} else |
53f9263ba mm: rework mapcou... |
3920 |
page_dup_rmap(page, true); |
1e8f889b1 [PATCH] Hugetlb: ... |
3921 3922 |
new_pte = make_huge_pte(vma, page, ((vma->vm_flags & VM_WRITE) && (vma->vm_flags & VM_SHARED))); |
285b8dcaa mm, hugetlbfs: pa... |
3923 |
set_huge_pte_at(mm, haddr, ptep, new_pte); |
1e8f889b1 [PATCH] Hugetlb: ... |
3924 |
|
5d317b2b6 mm: hugetlb: proc... |
3925 |
hugetlb_count_add(pages_per_huge_page(h), mm); |
788c7df45 hugetlb: fault fl... |
3926 |
if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) { |
1e8f889b1 [PATCH] Hugetlb: ... |
3927 |
/* Optimization, do the COW without a second fault */ |
974e6d66b mm, hugetlbfs: pa... |
3928 |
ret = hugetlb_cow(mm, vma, address, ptep, page, ptl); |
1e8f889b1 [PATCH] Hugetlb: ... |
3929 |
} |
cb900f412 mm, hugetlb: conv... |
3930 |
spin_unlock(ptl); |
cb6acd01e hugetlbfs: fix ra... |
3931 3932 3933 3934 3935 3936 3937 3938 |
/* * Only make newly allocated pages active. Existing pages found * in the pagecache could be !page_huge_active() if they have been * isolated for migration. */ if (new_page) set_page_huge_active(page); |
4c8872659 [PATCH] hugetlb: ... |
3939 3940 |
unlock_page(page); out: |
ac9b9c667 [PATCH] Fix handl... |
3941 |
return ret; |
4c8872659 [PATCH] hugetlb: ... |
3942 3943 |
backout: |
cb900f412 mm, hugetlb: conv... |
3944 |
spin_unlock(ptl); |
2b26736c8 allocate structur... |
3945 |
backout_unlocked: |
4c8872659 [PATCH] hugetlb: ... |
3946 |
unlock_page(page); |
285b8dcaa mm, hugetlbfs: pa... |
3947 |
restore_reserve_on_error(h, vma, haddr, page); |
4c8872659 [PATCH] hugetlb: ... |
3948 3949 |
put_page(page); goto out; |
ac9b9c667 [PATCH] Fix handl... |
3950 |
} |
8382d914e mm, hugetlb: impr... |
3951 |
#ifdef CONFIG_SMP |
1b426bac6 hugetlb: use same... |
3952 |
u32 hugetlb_fault_mutex_hash(struct hstate *h, struct address_space *mapping, |
8382d914e mm, hugetlb: impr... |
3953 3954 3955 3956 |
pgoff_t idx, unsigned long address) { unsigned long key[2]; u32 hash; |
1b426bac6 hugetlb: use same... |
3957 3958 |
key[0] = (unsigned long) mapping; key[1] = idx; |
8382d914e mm, hugetlb: impr... |
3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 |
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. */ |
1b426bac6 hugetlb: use same... |
3969 |
u32 hugetlb_fault_mutex_hash(struct hstate *h, struct address_space *mapping, |
8382d914e mm, hugetlb: impr... |
3970 3971 3972 3973 3974 |
pgoff_t idx, unsigned long address) { return 0; } #endif |
2b7403035 mm: Change return... |
3975 |
vm_fault_t hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
788c7df45 hugetlb: fault fl... |
3976 |
unsigned long address, unsigned int flags) |
86e5216f8 [PATCH] Hugetlb: ... |
3977 |
{ |
8382d914e mm, hugetlb: impr... |
3978 |
pte_t *ptep, entry; |
cb900f412 mm, hugetlb: conv... |
3979 |
spinlock_t *ptl; |
2b7403035 mm: Change return... |
3980 |
vm_fault_t ret; |
8382d914e mm, hugetlb: impr... |
3981 3982 |
u32 hash; pgoff_t idx; |
0fe6e20b9 hugetlb, rmap: ad... |
3983 |
struct page *page = NULL; |
57303d801 hugetlbfs: alloca... |
3984 |
struct page *pagecache_page = NULL; |
a55164389 hugetlb: modular ... |
3985 |
struct hstate *h = hstate_vma(vma); |
8382d914e mm, hugetlb: impr... |
3986 |
struct address_space *mapping; |
0f792cf94 mm/hugetlb: fix g... |
3987 |
int need_wait_lock = 0; |
285b8dcaa mm, hugetlbfs: pa... |
3988 |
unsigned long haddr = address & huge_page_mask(h); |
86e5216f8 [PATCH] Hugetlb: ... |
3989 |
|
285b8dcaa mm, hugetlbfs: pa... |
3990 |
ptep = huge_pte_offset(mm, haddr, huge_page_size(h)); |
fd6a03edd HWPOISON, hugetlb... |
3991 3992 |
if (ptep) { entry = huge_ptep_get(ptep); |
290408d4a hugetlb: hugepage... |
3993 |
if (unlikely(is_hugetlb_entry_migration(entry))) { |
cb900f412 mm, hugetlb: conv... |
3994 |
migration_entry_wait_huge(vma, mm, ptep); |
290408d4a hugetlb: hugepage... |
3995 3996 |
return 0; } else if (unlikely(is_hugetlb_entry_hwpoisoned(entry))) |
32f84528f mm: hugetlb: fix ... |
3997 |
return VM_FAULT_HWPOISON_LARGE | |
972dc4de1 hugetlb: add an i... |
3998 |
VM_FAULT_SET_HINDEX(hstate_index(h)); |
ddeaab32a hugetlbfs: revert... |
3999 4000 4001 4002 |
} else { ptep = huge_pte_alloc(mm, haddr, huge_page_size(h)); if (!ptep) return VM_FAULT_OOM; |
fd6a03edd HWPOISON, hugetlb... |
4003 |
} |
8382d914e mm, hugetlb: impr... |
4004 |
mapping = vma->vm_file->f_mapping; |
ddeaab32a hugetlbfs: revert... |
4005 |
idx = vma_hugecache_offset(h, vma, haddr); |
8382d914e mm, hugetlb: impr... |
4006 |
|
3935baa9b [PATCH] hugepage:... |
4007 4008 4009 4010 4011 |
/* * 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. */ |
1b426bac6 hugetlb: use same... |
4012 |
hash = hugetlb_fault_mutex_hash(h, mapping, idx, haddr); |
c672c7f29 mm/hugetlb: expos... |
4013 |
mutex_lock(&hugetlb_fault_mutex_table[hash]); |
8382d914e mm, hugetlb: impr... |
4014 |
|
7f2e9525b hugetlbfs: common... |
4015 4016 |
entry = huge_ptep_get(ptep); if (huge_pte_none(entry)) { |
8382d914e mm, hugetlb: impr... |
4017 |
ret = hugetlb_no_page(mm, vma, mapping, idx, address, ptep, flags); |
b4d1d99fd hugetlb: handle u... |
4018 |
goto out_mutex; |
3935baa9b [PATCH] hugepage:... |
4019 |
} |
86e5216f8 [PATCH] Hugetlb: ... |
4020 |
|
83c54070e mm: fault feedbac... |
4021 |
ret = 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
4022 |
|
57303d801 hugetlbfs: alloca... |
4023 |
/* |
0f792cf94 mm/hugetlb: fix g... |
4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 |
* entry could be a migration/hwpoison entry at this point, so this * check prevents the kernel from going below assuming that we have * a active hugepage in pagecache. This goto expects the 2nd page fault, * and is_hugetlb_entry_(migration|hwpoisoned) check will properly * handle it. */ if (!pte_present(entry)) goto out_mutex; /* |
57303d801 hugetlbfs: alloca... |
4034 4035 4036 4037 4038 4039 4040 |
* 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 mm/hugetlb: add m... |
4041 |
if ((flags & FAULT_FLAG_WRITE) && !huge_pte_write(entry)) { |
285b8dcaa mm, hugetlbfs: pa... |
4042 |
if (vma_needs_reservation(h, vma, haddr) < 0) { |
2b26736c8 allocate structur... |
4043 |
ret = VM_FAULT_OOM; |
b4d1d99fd hugetlb: handle u... |
4044 |
goto out_mutex; |
2b26736c8 allocate structur... |
4045 |
} |
5e9113731 mm/hugetlb: add c... |
4046 |
/* Just decrements count, does not deallocate */ |
285b8dcaa mm, hugetlbfs: pa... |
4047 |
vma_end_reservation(h, vma, haddr); |
57303d801 hugetlbfs: alloca... |
4048 |
|
f83a275db mm: account for M... |
4049 |
if (!(vma->vm_flags & VM_MAYSHARE)) |
57303d801 hugetlbfs: alloca... |
4050 |
pagecache_page = hugetlbfs_pagecache_page(h, |
285b8dcaa mm, hugetlbfs: pa... |
4051 |
vma, haddr); |
57303d801 hugetlbfs: alloca... |
4052 |
} |
0f792cf94 mm/hugetlb: fix g... |
4053 4054 4055 4056 4057 |
ptl = huge_pte_lock(h, mm, ptep); /* Check for a racing update before calling hugetlb_cow */ if (unlikely(!pte_same(entry, huge_ptep_get(ptep)))) goto out_ptl; |
56c9cfb13 hugetlb, rmap: fi... |
4058 4059 4060 4061 |
/* * 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. |
56c9cfb13 hugetlb, rmap: fi... |
4062 4063 4064 |
*/ page = pte_page(entry); if (page != pagecache_page) |
0f792cf94 mm/hugetlb: fix g... |
4065 4066 4067 4068 |
if (!trylock_page(page)) { need_wait_lock = 1; goto out_ptl; } |
b4d1d99fd hugetlb: handle u... |
4069 |
|
0f792cf94 mm/hugetlb: fix g... |
4070 |
get_page(page); |
b4d1d99fd hugetlb: handle u... |
4071 |
|
788c7df45 hugetlb: fault fl... |
4072 |
if (flags & FAULT_FLAG_WRITE) { |
106c992a5 mm/hugetlb: add m... |
4073 |
if (!huge_pte_write(entry)) { |
974e6d66b mm, hugetlbfs: pa... |
4074 |
ret = hugetlb_cow(mm, vma, address, ptep, |
3999f52e3 mm/hugetlb.c: use... |
4075 |
pagecache_page, ptl); |
0f792cf94 mm/hugetlb: fix g... |
4076 |
goto out_put_page; |
b4d1d99fd hugetlb: handle u... |
4077 |
} |
106c992a5 mm/hugetlb: add m... |
4078 |
entry = huge_pte_mkdirty(entry); |
b4d1d99fd hugetlb: handle u... |
4079 4080 |
} entry = pte_mkyoung(entry); |
285b8dcaa mm, hugetlbfs: pa... |
4081 |
if (huge_ptep_set_access_flags(vma, haddr, ptep, entry, |
788c7df45 hugetlb: fault fl... |
4082 |
flags & FAULT_FLAG_WRITE)) |
285b8dcaa mm, hugetlbfs: pa... |
4083 |
update_mmu_cache(vma, haddr, ptep); |
0f792cf94 mm/hugetlb: fix g... |
4084 4085 4086 4087 |
out_put_page: if (page != pagecache_page) unlock_page(page); put_page(page); |
cb900f412 mm, hugetlb: conv... |
4088 4089 |
out_ptl: spin_unlock(ptl); |
57303d801 hugetlbfs: alloca... |
4090 4091 4092 4093 4094 |
if (pagecache_page) { unlock_page(pagecache_page); put_page(pagecache_page); } |
b4d1d99fd hugetlb: handle u... |
4095 |
out_mutex: |
c672c7f29 mm/hugetlb: expos... |
4096 |
mutex_unlock(&hugetlb_fault_mutex_table[hash]); |
0f792cf94 mm/hugetlb: fix g... |
4097 4098 4099 4100 4101 4102 4103 4104 4105 |
/* * Generally it's safe to hold refcount during waiting page lock. But * here we just wait to defer the next page fault to avoid busy loop and * the page is not used after unlocked before returning from the current * page fault. So we are safe from accessing freed page, even if we wait * here without taking refcount. */ if (need_wait_lock) wait_on_page_locked(page); |
1e8f889b1 [PATCH] Hugetlb: ... |
4106 |
return ret; |
86e5216f8 [PATCH] Hugetlb: ... |
4107 |
} |
8fb5debc5 userfaultfd: huge... |
4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 |
/* * Used by userfaultfd UFFDIO_COPY. Based on mcopy_atomic_pte with * modifications for huge pages. */ int hugetlb_mcopy_atomic_pte(struct mm_struct *dst_mm, pte_t *dst_pte, struct vm_area_struct *dst_vma, unsigned long dst_addr, unsigned long src_addr, struct page **pagep) { |
1e3921471 userfaultfd: huge... |
4119 4120 4121 |
struct address_space *mapping; pgoff_t idx; unsigned long size; |
1c9e8def4 userfaultfd: huge... |
4122 |
int vm_shared = dst_vma->vm_flags & VM_SHARED; |
8fb5debc5 userfaultfd: huge... |
4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 |
struct hstate *h = hstate_vma(dst_vma); pte_t _dst_pte; spinlock_t *ptl; int ret; struct page *page; if (!*pagep) { ret = -ENOMEM; page = alloc_huge_page(dst_vma, dst_addr, 0); if (IS_ERR(page)) goto out; ret = copy_huge_page_from_user(page, (const void __user *) src_addr, |
810a56b94 userfaultfd: huge... |
4137 |
pages_per_huge_page(h), false); |
8fb5debc5 userfaultfd: huge... |
4138 4139 4140 |
/* fallback to copy_from_user outside mmap_sem */ if (unlikely(ret)) { |
9e368259a userfaultfd: use ... |
4141 |
ret = -ENOENT; |
8fb5debc5 userfaultfd: huge... |
4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 |
*pagep = page; /* don't free the page */ goto out; } } else { page = *pagep; *pagep = NULL; } /* * The memory barrier inside __SetPageUptodate makes sure that * preceding stores to the page contents become visible before * the set_pte_at() write. */ __SetPageUptodate(page); |
8fb5debc5 userfaultfd: huge... |
4157 |
|
1e3921471 userfaultfd: huge... |
4158 4159 |
mapping = dst_vma->vm_file->f_mapping; idx = vma_hugecache_offset(h, dst_vma, dst_addr); |
1c9e8def4 userfaultfd: huge... |
4160 4161 4162 4163 |
/* * If shared, add to page cache */ if (vm_shared) { |
1e3921471 userfaultfd: huge... |
4164 4165 4166 4167 |
size = i_size_read(mapping->host) >> huge_page_shift(h); ret = -EFAULT; if (idx >= size) goto out_release_nounlock; |
1c9e8def4 userfaultfd: huge... |
4168 |
|
1e3921471 userfaultfd: huge... |
4169 4170 4171 4172 4173 4174 |
/* * Serialization between remove_inode_hugepages() and * huge_add_to_page_cache() below happens through the * hugetlb_fault_mutex_table that here must be hold by * the caller. */ |
1c9e8def4 userfaultfd: huge... |
4175 4176 4177 4178 |
ret = huge_add_to_page_cache(page, mapping, idx); if (ret) goto out_release_nounlock; } |
8fb5debc5 userfaultfd: huge... |
4179 4180 |
ptl = huge_pte_lockptr(h, dst_mm, dst_pte); spin_lock(ptl); |
1e3921471 userfaultfd: huge... |
4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 |
/* * Recheck the i_size after holding PT lock to make sure not * to leave any page mapped (as page_mapped()) beyond the end * of the i_size (remove_inode_hugepages() is strict about * enforcing that). If we bail out here, we'll also leave a * page in the radix tree in the vm_shared case beyond the end * of the i_size, but remove_inode_hugepages() will take care * of it as soon as we drop the hugetlb_fault_mutex_table. */ size = i_size_read(mapping->host) >> huge_page_shift(h); ret = -EFAULT; if (idx >= size) goto out_release_unlock; |
8fb5debc5 userfaultfd: huge... |
4194 4195 4196 |
ret = -EEXIST; if (!huge_pte_none(huge_ptep_get(dst_pte))) goto out_release_unlock; |
1c9e8def4 userfaultfd: huge... |
4197 4198 4199 4200 4201 4202 |
if (vm_shared) { page_dup_rmap(page, true); } else { ClearPagePrivate(page); hugepage_add_new_anon_rmap(page, dst_vma, dst_addr); } |
8fb5debc5 userfaultfd: huge... |
4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 |
_dst_pte = make_huge_pte(dst_vma, page, dst_vma->vm_flags & VM_WRITE); if (dst_vma->vm_flags & VM_WRITE) _dst_pte = huge_pte_mkdirty(_dst_pte); _dst_pte = pte_mkyoung(_dst_pte); set_huge_pte_at(dst_mm, dst_addr, dst_pte, _dst_pte); (void)huge_ptep_set_access_flags(dst_vma, dst_addr, dst_pte, _dst_pte, dst_vma->vm_flags & VM_WRITE); hugetlb_count_add(pages_per_huge_page(h), dst_mm); /* No need to invalidate - it was non-present before */ update_mmu_cache(dst_vma, dst_addr, dst_pte); spin_unlock(ptl); |
cb6acd01e hugetlbfs: fix ra... |
4219 |
set_page_huge_active(page); |
1c9e8def4 userfaultfd: huge... |
4220 4221 |
if (vm_shared) unlock_page(page); |
8fb5debc5 userfaultfd: huge... |
4222 4223 4224 4225 4226 |
ret = 0; out: return ret; out_release_unlock: spin_unlock(ptl); |
1c9e8def4 userfaultfd: huge... |
4227 4228 |
if (vm_shared) unlock_page(page); |
5af10dfd0 userfaultfd: huge... |
4229 |
out_release_nounlock: |
8fb5debc5 userfaultfd: huge... |
4230 4231 4232 |
put_page(page); goto out; } |
28a35716d mm: use long type... |
4233 4234 4235 |
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, |
87ffc118b userfaultfd: huge... |
4236 |
long i, unsigned int flags, int *nonblocking) |
63551ae0f [PATCH] Hugepage ... |
4237 |
{ |
d5d4b0aa4 [PATCH] optimize ... |
4238 4239 |
unsigned long pfn_offset; unsigned long vaddr = *position; |
28a35716d mm: use long type... |
4240 |
unsigned long remainder = *nr_pages; |
a55164389 hugetlb: modular ... |
4241 |
struct hstate *h = hstate_vma(vma); |
2be7cfed9 mm/hugetlb.c: __g... |
4242 |
int err = -EFAULT; |
63551ae0f [PATCH] Hugepage ... |
4243 |
|
63551ae0f [PATCH] Hugepage ... |
4244 |
while (vaddr < vma->vm_end && remainder) { |
4c8872659 [PATCH] hugetlb: ... |
4245 |
pte_t *pte; |
cb900f412 mm, hugetlb: conv... |
4246 |
spinlock_t *ptl = NULL; |
2a15efc95 mm: follow_hugetl... |
4247 |
int absent; |
4c8872659 [PATCH] hugetlb: ... |
4248 |
struct page *page; |
63551ae0f [PATCH] Hugepage ... |
4249 |
|
4c8872659 [PATCH] hugetlb: ... |
4250 |
/* |
02057967b mm, hugetlb: abor... |
4251 4252 4253 |
* If we have a pending SIGKILL, don't keep faulting pages and * potentially allocating memory. */ |
fa45f1162 mm/: remove calle... |
4254 |
if (fatal_signal_pending(current)) { |
02057967b mm, hugetlb: abor... |
4255 4256 4257 4258 4259 |
remainder = 0; break; } /* |
4c8872659 [PATCH] hugetlb: ... |
4260 |
* Some archs (sparc64, sh*) have multiple pte_ts to |
2a15efc95 mm: follow_hugetl... |
4261 |
* each hugepage. We have to make sure we get the |
4c8872659 [PATCH] hugetlb: ... |
4262 |
* first, for the page indexing below to work. |
cb900f412 mm, hugetlb: conv... |
4263 4264 |
* * Note that page table lock is not held when pte is null. |
4c8872659 [PATCH] hugetlb: ... |
4265 |
*/ |
7868a2087 mm/hugetlb: add s... |
4266 4267 |
pte = huge_pte_offset(mm, vaddr & huge_page_mask(h), huge_page_size(h)); |
cb900f412 mm, hugetlb: conv... |
4268 4269 |
if (pte) ptl = huge_pte_lock(h, mm, pte); |
2a15efc95 mm: follow_hugetl... |
4270 4271 4272 4273 |
absent = !pte || huge_pte_none(huge_ptep_get(pte)); /* * When coredumping, it suits get_dump_page if we just return |
3ae77f43b mm: hugetlbfs_pag... |
4274 4275 4276 4277 |
* 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 mm: follow_hugetl... |
4278 |
*/ |
3ae77f43b mm: hugetlbfs_pag... |
4279 4280 |
if (absent && (flags & FOLL_DUMP) && !hugetlbfs_pagecache_present(h, vma, vaddr)) { |
cb900f412 mm, hugetlb: conv... |
4281 4282 |
if (pte) spin_unlock(ptl); |
2a15efc95 mm: follow_hugetl... |
4283 4284 4285 |
remainder = 0; break; } |
63551ae0f [PATCH] Hugepage ... |
4286 |
|
9cc3a5bd4 hugetlbfs: add sw... |
4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 |
/* * 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 mm/hugetlb: add m... |
4298 4299 |
((flags & FOLL_WRITE) && !huge_pte_write(huge_ptep_get(pte)))) { |
2b7403035 mm: Change return... |
4300 |
vm_fault_t ret; |
87ffc118b userfaultfd: huge... |
4301 |
unsigned int fault_flags = 0; |
63551ae0f [PATCH] Hugepage ... |
4302 |
|
cb900f412 mm, hugetlb: conv... |
4303 4304 |
if (pte) spin_unlock(ptl); |
87ffc118b userfaultfd: huge... |
4305 4306 4307 4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 |
if (flags & FOLL_WRITE) fault_flags |= FAULT_FLAG_WRITE; if (nonblocking) fault_flags |= FAULT_FLAG_ALLOW_RETRY; if (flags & FOLL_NOWAIT) fault_flags |= FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_RETRY_NOWAIT; if (flags & FOLL_TRIED) { VM_WARN_ON_ONCE(fault_flags & FAULT_FLAG_ALLOW_RETRY); fault_flags |= FAULT_FLAG_TRIED; } ret = hugetlb_fault(mm, vma, vaddr, fault_flags); if (ret & VM_FAULT_ERROR) { |
2be7cfed9 mm/hugetlb.c: __g... |
4319 |
err = vm_fault_to_errno(ret, flags); |
87ffc118b userfaultfd: huge... |
4320 4321 4322 4323 |
remainder = 0; break; } if (ret & VM_FAULT_RETRY) { |
1ac25013f mm/hugetlb.c: tea... |
4324 4325 |
if (nonblocking && !(fault_flags & FAULT_FLAG_RETRY_NOWAIT)) |
87ffc118b userfaultfd: huge... |
4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 |
*nonblocking = 0; *nr_pages = 0; /* * VM_FAULT_RETRY must not return an * error, it will return zero * instead. * * No need to update "position" as the * caller will not check it after * *nr_pages is set to 0. */ return i; } continue; |
4c8872659 [PATCH] hugetlb: ... |
4340 |
} |
a55164389 hugetlb: modular ... |
4341 |
pfn_offset = (vaddr & ~huge_page_mask(h)) >> PAGE_SHIFT; |
7f2e9525b hugetlbfs: common... |
4342 |
page = pte_page(huge_ptep_get(pte)); |
8fde12ca7 mm: prevent get_u... |
4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 |
/* * Instead of doing 'try_get_page()' below in the same_page * loop, just check the count once here. */ if (unlikely(page_count(page) <= 0)) { if (pages) { spin_unlock(ptl); remainder = 0; err = -ENOMEM; break; } } |
d5d4b0aa4 [PATCH] optimize ... |
4356 |
same_page: |
d6692183a [PATCH] fix extra... |
4357 |
if (pages) { |
2a15efc95 mm: follow_hugetl... |
4358 |
pages[i] = mem_map_offset(page, pfn_offset); |
ddc58f27f mm: drop tail pag... |
4359 |
get_page(pages[i]); |
d6692183a [PATCH] fix extra... |
4360 |
} |
63551ae0f [PATCH] Hugepage ... |
4361 4362 4363 4364 4365 |
if (vmas) vmas[i] = vma; vaddr += PAGE_SIZE; |
d5d4b0aa4 [PATCH] optimize ... |
4366 |
++pfn_offset; |
63551ae0f [PATCH] Hugepage ... |
4367 4368 |
--remainder; ++i; |
d5d4b0aa4 [PATCH] optimize ... |
4369 |
if (vaddr < vma->vm_end && remainder && |
a55164389 hugetlb: modular ... |
4370 |
pfn_offset < pages_per_huge_page(h)) { |
d5d4b0aa4 [PATCH] optimize ... |
4371 4372 4373 4374 4375 4376 |
/* * We use pfn_offset to avoid touching the pageframes * of this compound page. */ goto same_page; } |
cb900f412 mm, hugetlb: conv... |
4377 |
spin_unlock(ptl); |
63551ae0f [PATCH] Hugepage ... |
4378 |
} |
28a35716d mm: use long type... |
4379 |
*nr_pages = remainder; |
87ffc118b userfaultfd: huge... |
4380 4381 4382 4383 4384 |
/* * setting position is actually required only if remainder is * not zero but it's faster not to add a "if (remainder)" * branch. */ |
63551ae0f [PATCH] Hugepage ... |
4385 |
*position = vaddr; |
2be7cfed9 mm/hugetlb.c: __g... |
4386 |
return i ? i : err; |
63551ae0f [PATCH] Hugepage ... |
4387 |
} |
8f860591f [PATCH] Enable mp... |
4388 |
|
5491ae7b6 powerpc/mm/hugetl... |
4389 4390 4391 4392 4393 4394 4395 |
#ifndef __HAVE_ARCH_FLUSH_HUGETLB_TLB_RANGE /* * ARCHes with special requirements for evicting HUGETLB backing TLB entries can * implement this. */ #define flush_hugetlb_tlb_range(vma, addr, end) flush_tlb_range(vma, addr, end) #endif |
7da4d641c mm: Count the num... |
4396 |
unsigned long hugetlb_change_protection(struct vm_area_struct *vma, |
8f860591f [PATCH] Enable mp... |
4397 4398 4399 4400 4401 4402 |
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 hugetlb: modular ... |
4403 |
struct hstate *h = hstate_vma(vma); |
7da4d641c mm: Count the num... |
4404 |
unsigned long pages = 0; |
dff11abe2 hugetlb: take PMD... |
4405 |
bool shared_pmd = false; |
ac46d4f3c mm/mmu_notifier: ... |
4406 |
struct mmu_notifier_range range; |
dff11abe2 hugetlb: take PMD... |
4407 4408 4409 |
/* * In the case of shared PMDs, the area to flush could be beyond |
ac46d4f3c mm/mmu_notifier: ... |
4410 |
* start/end. Set range.start/range.end to cover the maximum possible |
dff11abe2 hugetlb: take PMD... |
4411 4412 |
* range if PMD sharing is possible. */ |
7269f9999 mm/mmu_notifier: ... |
4413 4414 |
mmu_notifier_range_init(&range, MMU_NOTIFY_PROTECTION_VMA, 0, vma, mm, start, end); |
ac46d4f3c mm/mmu_notifier: ... |
4415 |
adjust_range_if_pmd_sharing_possible(vma, &range.start, &range.end); |
8f860591f [PATCH] Enable mp... |
4416 4417 |
BUG_ON(address >= end); |
ac46d4f3c mm/mmu_notifier: ... |
4418 |
flush_cache_range(vma, range.start, range.end); |
8f860591f [PATCH] Enable mp... |
4419 |
|
ac46d4f3c mm/mmu_notifier: ... |
4420 |
mmu_notifier_invalidate_range_start(&range); |
83cde9e8b mm: use new helpe... |
4421 |
i_mmap_lock_write(vma->vm_file->f_mapping); |
a55164389 hugetlb: modular ... |
4422 |
for (; address < end; address += huge_page_size(h)) { |
cb900f412 mm, hugetlb: conv... |
4423 |
spinlock_t *ptl; |
7868a2087 mm/hugetlb: add s... |
4424 |
ptep = huge_pte_offset(mm, address, huge_page_size(h)); |
8f860591f [PATCH] Enable mp... |
4425 4426 |
if (!ptep) continue; |
cb900f412 mm, hugetlb: conv... |
4427 |
ptl = huge_pte_lock(h, mm, ptep); |
7da4d641c mm: Count the num... |
4428 4429 |
if (huge_pmd_unshare(mm, &address, ptep)) { pages++; |
cb900f412 mm, hugetlb: conv... |
4430 |
spin_unlock(ptl); |
dff11abe2 hugetlb: take PMD... |
4431 |
shared_pmd = true; |
39dde65c9 [PATCH] shared pa... |
4432 |
continue; |
7da4d641c mm: Count the num... |
4433 |
} |
a8bda28d8 mm/hugetlb: add m... |
4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 |
pte = huge_ptep_get(ptep); if (unlikely(is_hugetlb_entry_hwpoisoned(pte))) { spin_unlock(ptl); continue; } if (unlikely(is_hugetlb_entry_migration(pte))) { swp_entry_t entry = pte_to_swp_entry(pte); if (is_write_migration_entry(entry)) { pte_t newpte; make_migration_entry_read(&entry); newpte = swp_entry_to_pte(entry); |
e5251fd43 mm/hugetlb: intro... |
4447 4448 |
set_huge_swap_pte_at(mm, address, ptep, newpte, huge_page_size(h)); |
a8bda28d8 mm/hugetlb: add m... |
4449 4450 4451 4452 4453 4454 |
pages++; } spin_unlock(ptl); continue; } if (!huge_pte_none(pte)) { |
023bdd002 mm/hugetlb: add p... |
4455 4456 4457 4458 |
pte_t old_pte; old_pte = huge_ptep_modify_prot_start(vma, address, ptep); pte = pte_mkhuge(huge_pte_modify(old_pte, newprot)); |
be7517d6a mm/hugetlb: set P... |
4459 |
pte = arch_make_huge_pte(pte, vma, NULL, 0); |
023bdd002 mm/hugetlb: add p... |
4460 |
huge_ptep_modify_prot_commit(vma, address, ptep, old_pte, pte); |
7da4d641c mm: Count the num... |
4461 |
pages++; |
8f860591f [PATCH] Enable mp... |
4462 |
} |
cb900f412 mm, hugetlb: conv... |
4463 |
spin_unlock(ptl); |
8f860591f [PATCH] Enable mp... |
4464 |
} |
d833352a4 mm: hugetlbfs: cl... |
4465 |
/* |
c8c06efa8 mm: convert i_mma... |
4466 |
* Must flush TLB before releasing i_mmap_rwsem: x86's huge_pmd_unshare |
d833352a4 mm: hugetlbfs: cl... |
4467 |
* may have cleared our pud entry and done put_page on the page table: |
c8c06efa8 mm: convert i_mma... |
4468 |
* once we release i_mmap_rwsem, another task can do the final put_page |
dff11abe2 hugetlb: take PMD... |
4469 4470 |
* and that page table be reused and filled with junk. If we actually * did unshare a page of pmds, flush the range corresponding to the pud. |
d833352a4 mm: hugetlbfs: cl... |
4471 |
*/ |
dff11abe2 hugetlb: take PMD... |
4472 |
if (shared_pmd) |
ac46d4f3c mm/mmu_notifier: ... |
4473 |
flush_hugetlb_tlb_range(vma, range.start, range.end); |
dff11abe2 hugetlb: take PMD... |
4474 4475 |
else flush_hugetlb_tlb_range(vma, start, end); |
0f10851ea mm/mmu_notifier: ... |
4476 4477 4478 4479 |
/* * No need to call mmu_notifier_invalidate_range() we are downgrading * page table protection not changing it to point to a new page. * |
ad56b738c docs/vm: rename d... |
4480 |
* See Documentation/vm/mmu_notifier.rst |
0f10851ea mm/mmu_notifier: ... |
4481 |
*/ |
83cde9e8b mm: use new helpe... |
4482 |
i_mmap_unlock_write(vma->vm_file->f_mapping); |
ac46d4f3c mm/mmu_notifier: ... |
4483 |
mmu_notifier_invalidate_range_end(&range); |
7da4d641c mm: Count the num... |
4484 4485 |
return pages << h->order; |
8f860591f [PATCH] Enable mp... |
4486 |
} |
a1e78772d hugetlb: reserve ... |
4487 4488 |
int hugetlb_reserve_pages(struct inode *inode, long from, long to, |
5a6fe1259 Do not account fo... |
4489 |
struct vm_area_struct *vma, |
ca16d140a mm: don't access ... |
4490 |
vm_flags_t vm_flags) |
e4e574b76 hugetlb: Try to g... |
4491 |
{ |
17c9d12e1 Do not account fo... |
4492 |
long ret, chg; |
a55164389 hugetlb: modular ... |
4493 |
struct hstate *h = hstate_inode(inode); |
90481622d hugepages: fix us... |
4494 |
struct hugepage_subpool *spool = subpool_inode(inode); |
9119a41e9 mm, hugetlb: unif... |
4495 |
struct resv_map *resv_map; |
1c5ecae3a hugetlbfs: add mi... |
4496 |
long gbl_reserve; |
e4e574b76 hugetlb: Try to g... |
4497 |
|
63489f8e8 hugetlbfs: check ... |
4498 4499 4500 4501 4502 4503 |
/* This should never happen */ if (from > to) { VM_WARN(1, "%s called with a negative range ", __func__); return -EINVAL; } |
a1e78772d hugetlb: reserve ... |
4504 |
/* |
17c9d12e1 Do not account fo... |
4505 4506 |
* Only apply hugepage reservation if asked. At fault time, an * attempt will be made for VM_NORESERVE to allocate a page |
90481622d hugepages: fix us... |
4507 |
* without using reserves |
17c9d12e1 Do not account fo... |
4508 |
*/ |
ca16d140a mm: don't access ... |
4509 |
if (vm_flags & VM_NORESERVE) |
17c9d12e1 Do not account fo... |
4510 4511 4512 |
return 0; /* |
a1e78772d hugetlb: reserve ... |
4513 4514 4515 4516 4517 |
* 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 mm, hugetlb: unif... |
4518 |
if (!vma || vma->vm_flags & VM_MAYSHARE) { |
f27a5136f hugetlbfs: always... |
4519 4520 4521 4522 4523 |
/* * resv_map can not be NULL as hugetlb_reserve_pages is only * called for inodes for which resv_maps were created (see * hugetlbfs_get_inode). */ |
4e35f4838 mm, hugetlb: use ... |
4524 |
resv_map = inode_resv_map(inode); |
9119a41e9 mm, hugetlb: unif... |
4525 |
|
1406ec9ba mm, hugetlb: impr... |
4526 |
chg = region_chg(resv_map, from, to); |
9119a41e9 mm, hugetlb: unif... |
4527 4528 4529 |
} else { resv_map = resv_map_alloc(); |
17c9d12e1 Do not account fo... |
4530 4531 |
if (!resv_map) return -ENOMEM; |
a1e78772d hugetlb: reserve ... |
4532 |
chg = to - from; |
84afd99b8 hugetlb reservati... |
4533 |
|
17c9d12e1 Do not account fo... |
4534 4535 4536 |
set_vma_resv_map(vma, resv_map); set_vma_resv_flags(vma, HPAGE_RESV_OWNER); } |
c50ac0508 hugetlb: fix resv... |
4537 4538 4539 4540 |
if (chg < 0) { ret = chg; goto out_err; } |
8a6301127 pretend cpuset ha... |
4541 |
|
1c5ecae3a hugetlbfs: add mi... |
4542 4543 4544 4545 4546 4547 4548 |
/* * There must be enough pages in the subpool for the mapping. If * the subpool has a minimum size, there may be some global * reservations already in place (gbl_reserve). */ gbl_reserve = hugepage_subpool_get_pages(spool, chg); if (gbl_reserve < 0) { |
c50ac0508 hugetlb: fix resv... |
4549 4550 4551 |
ret = -ENOSPC; goto out_err; } |
5a6fe1259 Do not account fo... |
4552 4553 |
/* |
17c9d12e1 Do not account fo... |
4554 |
* Check enough hugepages are available for the reservation. |
90481622d hugepages: fix us... |
4555 |
* Hand the pages back to the subpool if there are not |
5a6fe1259 Do not account fo... |
4556 |
*/ |
1c5ecae3a hugetlbfs: add mi... |
4557 |
ret = hugetlb_acct_memory(h, gbl_reserve); |
68842c9b9 hugetlbfs: fix qu... |
4558 |
if (ret < 0) { |
1c5ecae3a hugetlbfs: add mi... |
4559 4560 |
/* put back original number of pages, chg */ (void)hugepage_subpool_put_pages(spool, chg); |
c50ac0508 hugetlb: fix resv... |
4561 |
goto out_err; |
68842c9b9 hugetlbfs: fix qu... |
4562 |
} |
17c9d12e1 Do not account fo... |
4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 |
/* * 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 */ |
33039678c mm/hugetlb: handl... |
4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 |
if (!vma || vma->vm_flags & VM_MAYSHARE) { long add = region_add(resv_map, from, to); if (unlikely(chg > add)) { /* * pages in this range were added to the reserve * map between region_chg and region_add. This * indicates a race with alloc_huge_page. Adjust * the subpool and reserve counts modified above * based on the difference. */ long rsv_adjust; rsv_adjust = hugepage_subpool_put_pages(spool, chg - add); hugetlb_acct_memory(h, -rsv_adjust); } } |
a43a8c39b [PATCH] tightenin... |
4593 |
return 0; |
c50ac0508 hugetlb: fix resv... |
4594 |
out_err: |
5e9113731 mm/hugetlb: add c... |
4595 |
if (!vma || vma->vm_flags & VM_MAYSHARE) |
ff8c0c53c mm/hugetlb.c: don... |
4596 4597 4598 |
/* Don't call region_abort if region_chg failed */ if (chg >= 0) region_abort(resv_map, from, to); |
f031dd274 mm, hugetlb: remo... |
4599 4600 |
if (vma && is_vma_resv_set(vma, HPAGE_RESV_OWNER)) kref_put(&resv_map->refs, resv_map_release); |
c50ac0508 hugetlb: fix resv... |
4601 |
return ret; |
a43a8c39b [PATCH] tightenin... |
4602 |
} |
b5cec28d3 hugetlbfs: trunca... |
4603 4604 |
long hugetlb_unreserve_pages(struct inode *inode, long start, long end, long freed) |
a43a8c39b [PATCH] tightenin... |
4605 |
{ |
a55164389 hugetlb: modular ... |
4606 |
struct hstate *h = hstate_inode(inode); |
4e35f4838 mm, hugetlb: use ... |
4607 |
struct resv_map *resv_map = inode_resv_map(inode); |
9119a41e9 mm, hugetlb: unif... |
4608 |
long chg = 0; |
90481622d hugepages: fix us... |
4609 |
struct hugepage_subpool *spool = subpool_inode(inode); |
1c5ecae3a hugetlbfs: add mi... |
4610 |
long gbl_reserve; |
45c682a68 hugetlb: fix i_bl... |
4611 |
|
f27a5136f hugetlbfs: always... |
4612 4613 4614 4615 |
/* * Since this routine can be called in the evict inode path for all * hugetlbfs inodes, resv_map could be NULL. */ |
b5cec28d3 hugetlbfs: trunca... |
4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 |
if (resv_map) { chg = region_del(resv_map, start, end); /* * region_del() can fail in the rare case where a region * must be split and another region descriptor can not be * allocated. If end == LONG_MAX, it will not fail. */ if (chg < 0) return chg; } |
45c682a68 hugetlb: fix i_bl... |
4626 |
spin_lock(&inode->i_lock); |
e4c6f8bed hugetlbfs: fix i_... |
4627 |
inode->i_blocks -= (blocks_per_huge_page(h) * freed); |
45c682a68 hugetlb: fix i_bl... |
4628 |
spin_unlock(&inode->i_lock); |
1c5ecae3a hugetlbfs: add mi... |
4629 4630 4631 4632 4633 4634 |
/* * If the subpool has a minimum size, the number of global * reservations to be released may be adjusted. */ gbl_reserve = hugepage_subpool_put_pages(spool, (chg - freed)); hugetlb_acct_memory(h, -gbl_reserve); |
b5cec28d3 hugetlbfs: trunca... |
4635 4636 |
return 0; |
a43a8c39b [PATCH] tightenin... |
4637 |
} |
93f70f900 HWPOISON, hugetlb... |
4638 |
|
3212b535f mm: hugetlb: Copy... |
4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 |
#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 */ |
de60f5f10 mm: introduce VM_... |
4650 4651 |
unsigned long vm_flags = vma->vm_flags & VM_LOCKED_CLEAR_MASK; unsigned long svm_flags = svma->vm_flags & VM_LOCKED_CLEAR_MASK; |
3212b535f mm: hugetlb: Copy... |
4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 |
/* * 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; } |
31aafb45f mm/hugetlb.c: mak... |
4664 |
static bool vma_shareable(struct vm_area_struct *vma, unsigned long addr) |
3212b535f mm: hugetlb: Copy... |
4665 4666 4667 4668 4669 4670 4671 |
{ unsigned long base = addr & PUD_MASK; unsigned long end = base + PUD_SIZE; /* * check on proper vm_flags and page table alignment */ |
017b1660d mm: migration: fi... |
4672 |
if (vma->vm_flags & VM_MAYSHARE && range_in_vma(vma, base, end)) |
31aafb45f mm/hugetlb.c: mak... |
4673 4674 |
return true; return false; |
3212b535f mm: hugetlb: Copy... |
4675 4676 4677 |
} /* |
017b1660d mm: migration: fi... |
4678 4679 4680 4681 4682 4683 4684 |
* Determine if start,end range within vma could be mapped by shared pmd. * If yes, adjust start and end to cover range associated with possible * shared pmd mappings. */ void adjust_range_if_pmd_sharing_possible(struct vm_area_struct *vma, unsigned long *start, unsigned long *end) { |
d6bca2a8f mm/hugetlb: fix c... |
4685 |
unsigned long a_start, a_end; |
017b1660d mm: migration: fi... |
4686 4687 4688 |
if (!(vma->vm_flags & VM_MAYSHARE)) return; |
d6bca2a8f mm/hugetlb: fix c... |
4689 4690 4691 |
/* Extend the range to be PUD aligned for a worst case scenario */ a_start = ALIGN_DOWN(*start, PUD_SIZE); a_end = ALIGN(*end, PUD_SIZE); |
017b1660d mm: migration: fi... |
4692 |
|
d6bca2a8f mm/hugetlb: fix c... |
4693 4694 4695 4696 4697 4698 |
/* * Intersect the range with the vma range, since pmd sharing won't be * across vma after all */ *start = max(vma->vm_start, a_start); *end = min(vma->vm_end, a_end); |
017b1660d mm: migration: fi... |
4699 4700 4701 |
} /* |
3212b535f mm: hugetlb: Copy... |
4702 4703 4704 |
* 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 |
ddeaab32a hugetlbfs: revert... |
4705 4706 4707 4708 |
* code much cleaner. pmd allocation is essential for the shared case because * pud has to be populated inside the same i_mmap_rwsem section - otherwise * racing tasks could either miss the sharing (see huge_pte_offset) or select a * bad pmd for sharing. |
3212b535f mm: hugetlb: Copy... |
4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 |
*/ 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 mm, hugetlb: conv... |
4720 |
spinlock_t *ptl; |
3212b535f mm: hugetlb: Copy... |
4721 4722 4723 |
if (!vma_shareable(vma, addr)) return (pte_t *)pmd_alloc(mm, pud, addr); |
ddeaab32a hugetlbfs: revert... |
4724 |
i_mmap_lock_write(mapping); |
3212b535f mm: hugetlb: Copy... |
4725 4726 4727 4728 4729 4730 |
vma_interval_tree_foreach(svma, &mapping->i_mmap, idx, idx) { if (svma == vma) continue; saddr = page_table_shareable(svma, vma, addr, idx); if (saddr) { |
7868a2087 mm/hugetlb: add s... |
4731 4732 |
spte = huge_pte_offset(svma->vm_mm, saddr, vma_mmu_pagesize(svma)); |
3212b535f mm: hugetlb: Copy... |
4733 4734 4735 4736 4737 4738 4739 4740 4741 |
if (spte) { get_page(virt_to_page(spte)); break; } } } if (!spte) goto out; |
8bea80520 mm/hugetlb.c: use... |
4742 |
ptl = huge_pte_lock(hstate_vma(vma), mm, spte); |
dc6c9a35b mm: account pmd p... |
4743 |
if (pud_none(*pud)) { |
3212b535f mm: hugetlb: Copy... |
4744 4745 |
pud_populate(mm, pud, (pmd_t *)((unsigned long)spte & PAGE_MASK)); |
c17b1f425 hugetlb: fix nr_p... |
4746 |
mm_inc_nr_pmds(mm); |
dc6c9a35b mm: account pmd p... |
4747 |
} else { |
3212b535f mm: hugetlb: Copy... |
4748 |
put_page(virt_to_page(spte)); |
dc6c9a35b mm: account pmd p... |
4749 |
} |
cb900f412 mm, hugetlb: conv... |
4750 |
spin_unlock(ptl); |
3212b535f mm: hugetlb: Copy... |
4751 4752 |
out: pte = (pte_t *)pmd_alloc(mm, pud, addr); |
ddeaab32a hugetlbfs: revert... |
4753 |
i_mmap_unlock_write(mapping); |
3212b535f mm: hugetlb: Copy... |
4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 |
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. * |
ddeaab32a hugetlbfs: revert... |
4764 |
* called with page table lock held. |
3212b535f mm: hugetlb: Copy... |
4765 4766 4767 4768 4769 4770 4771 |
* * 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); |
c2febafc6 mm: convert gener... |
4772 4773 |
p4d_t *p4d = p4d_offset(pgd, *addr); pud_t *pud = pud_offset(p4d, *addr); |
3212b535f mm: hugetlb: Copy... |
4774 4775 4776 4777 4778 4779 4780 |
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)); |
dc6c9a35b mm: account pmd p... |
4781 |
mm_dec_nr_pmds(mm); |
3212b535f mm: hugetlb: Copy... |
4782 4783 4784 |
*addr = ALIGN(*addr, HPAGE_SIZE * PTRS_PER_PTE) - HPAGE_SIZE; return 1; } |
9e5fc74c3 mm: hugetlb: Copy... |
4785 4786 4787 4788 4789 4790 |
#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; } |
e81f2d223 mm/hugetlb: reduc... |
4791 4792 4793 4794 4795 |
int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep) { return 0; } |
017b1660d mm: migration: fi... |
4796 4797 4798 4799 4800 |
void adjust_range_if_pmd_sharing_possible(struct vm_area_struct *vma, unsigned long *start, unsigned long *end) { } |
9e5fc74c3 mm: hugetlb: Copy... |
4801 |
#define want_pmd_share() (0) |
3212b535f mm: hugetlb: Copy... |
4802 |
#endif /* CONFIG_ARCH_WANT_HUGE_PMD_SHARE */ |
9e5fc74c3 mm: hugetlb: Copy... |
4803 4804 4805 4806 4807 |
#ifdef CONFIG_ARCH_WANT_GENERAL_HUGETLB pte_t *huge_pte_alloc(struct mm_struct *mm, unsigned long addr, unsigned long sz) { pgd_t *pgd; |
c2febafc6 mm: convert gener... |
4808 |
p4d_t *p4d; |
9e5fc74c3 mm: hugetlb: Copy... |
4809 4810 4811 4812 |
pud_t *pud; pte_t *pte = NULL; pgd = pgd_offset(mm, addr); |
f4f0a3d85 mm/hugetlb: fix N... |
4813 4814 4815 |
p4d = p4d_alloc(mm, pgd, addr); if (!p4d) return NULL; |
c2febafc6 mm: convert gener... |
4816 |
pud = pud_alloc(mm, p4d, addr); |
9e5fc74c3 mm: hugetlb: Copy... |
4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 |
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); } } |
4e666314d mm, hugetlb: fix ... |
4828 |
BUG_ON(pte && pte_present(*pte) && !pte_huge(*pte)); |
9e5fc74c3 mm: hugetlb: Copy... |
4829 4830 4831 |
return pte; } |
9b19df292 mm/hugetlb.c: mak... |
4832 4833 4834 4835 4836 4837 4838 4839 4840 |
/* * huge_pte_offset() - Walk the page table to resolve the hugepage * entry at address @addr * * Return: Pointer to page table or swap entry (PUD or PMD) for * address @addr, or NULL if a p*d_none() entry is encountered and the * size @sz doesn't match the hugepage size at this level of the page * table. */ |
7868a2087 mm/hugetlb: add s... |
4841 4842 |
pte_t *huge_pte_offset(struct mm_struct *mm, unsigned long addr, unsigned long sz) |
9e5fc74c3 mm: hugetlb: Copy... |
4843 4844 |
{ pgd_t *pgd; |
c2febafc6 mm: convert gener... |
4845 |
p4d_t *p4d; |
3c88e95cd mm/hugetlb: fix a... |
4846 4847 |
pud_t *pud, pud_entry; pmd_t *pmd, pmd_entry; |
9e5fc74c3 mm: hugetlb: Copy... |
4848 4849 |
pgd = pgd_offset(mm, addr); |
c2febafc6 mm: convert gener... |
4850 4851 4852 4853 4854 |
if (!pgd_present(*pgd)) return NULL; p4d = p4d_offset(pgd, addr); if (!p4d_present(*p4d)) return NULL; |
9b19df292 mm/hugetlb.c: mak... |
4855 |
|
c2febafc6 mm: convert gener... |
4856 |
pud = pud_offset(p4d, addr); |
3c88e95cd mm/hugetlb: fix a... |
4857 4858 |
pud_entry = READ_ONCE(*pud); if (sz != PUD_SIZE && pud_none(pud_entry)) |
c2febafc6 mm: convert gener... |
4859 |
return NULL; |
9b19df292 mm/hugetlb.c: mak... |
4860 |
/* hugepage or swap? */ |
3c88e95cd mm/hugetlb: fix a... |
4861 |
if (pud_huge(pud_entry) || !pud_present(pud_entry)) |
c2febafc6 mm: convert gener... |
4862 |
return (pte_t *)pud; |
9b19df292 mm/hugetlb.c: mak... |
4863 |
|
c2febafc6 mm: convert gener... |
4864 |
pmd = pmd_offset(pud, addr); |
3c88e95cd mm/hugetlb: fix a... |
4865 4866 |
pmd_entry = READ_ONCE(*pmd); if (sz != PMD_SIZE && pmd_none(pmd_entry)) |
9b19df292 mm/hugetlb.c: mak... |
4867 4868 |
return NULL; /* hugepage or swap? */ |
3c88e95cd mm/hugetlb: fix a... |
4869 |
if (pmd_huge(pmd_entry) || !pmd_present(pmd_entry)) |
9b19df292 mm/hugetlb.c: mak... |
4870 4871 4872 |
return (pte_t *)pmd; return NULL; |
9e5fc74c3 mm: hugetlb: Copy... |
4873 |
} |
61f77eda9 mm/hugetlb: reduc... |
4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887 |
#endif /* CONFIG_ARCH_WANT_GENERAL_HUGETLB */ /* * These functions are overwritable if your architecture needs its own * behavior. */ struct page * __weak follow_huge_addr(struct mm_struct *mm, unsigned long address, int write) { return ERR_PTR(-EINVAL); } struct page * __weak |
4dc71451a mm/follow_page_ma... |
4888 4889 4890 4891 4892 4893 4894 4895 4896 |
follow_huge_pd(struct vm_area_struct *vma, unsigned long address, hugepd_t hpd, int flags, int pdshift) { WARN(1, "hugepd follow called with no support for hugepage directory format "); return NULL; } struct page * __weak |
9e5fc74c3 mm: hugetlb: Copy... |
4897 |
follow_huge_pmd(struct mm_struct *mm, unsigned long address, |
e66f17ff7 mm/hugetlb: take ... |
4898 |
pmd_t *pmd, int flags) |
9e5fc74c3 mm: hugetlb: Copy... |
4899 |
{ |
e66f17ff7 mm/hugetlb: take ... |
4900 4901 |
struct page *page = NULL; spinlock_t *ptl; |
c9d398fa2 mm, hugetlb: use ... |
4902 |
pte_t pte; |
e66f17ff7 mm/hugetlb: take ... |
4903 4904 4905 4906 4907 4908 4909 4910 4911 |
retry: ptl = pmd_lockptr(mm, pmd); spin_lock(ptl); /* * make sure that the address range covered by this pmd is not * unmapped from other threads. */ if (!pmd_huge(*pmd)) goto out; |
c9d398fa2 mm, hugetlb: use ... |
4912 4913 |
pte = huge_ptep_get((pte_t *)pmd); if (pte_present(pte)) { |
975341270 mm/hugetlb: use p... |
4914 |
page = pmd_page(*pmd) + ((address & ~PMD_MASK) >> PAGE_SHIFT); |
e66f17ff7 mm/hugetlb: take ... |
4915 4916 4917 |
if (flags & FOLL_GET) get_page(page); } else { |
c9d398fa2 mm, hugetlb: use ... |
4918 |
if (is_hugetlb_entry_migration(pte)) { |
e66f17ff7 mm/hugetlb: take ... |
4919 4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 |
spin_unlock(ptl); __migration_entry_wait(mm, (pte_t *)pmd, ptl); goto retry; } /* * hwpoisoned entry is treated as no_page_table in * follow_page_mask(). */ } out: spin_unlock(ptl); |
9e5fc74c3 mm: hugetlb: Copy... |
4930 4931 |
return page; } |
61f77eda9 mm/hugetlb: reduc... |
4932 |
struct page * __weak |
9e5fc74c3 mm: hugetlb: Copy... |
4933 |
follow_huge_pud(struct mm_struct *mm, unsigned long address, |
e66f17ff7 mm/hugetlb: take ... |
4934 |
pud_t *pud, int flags) |
9e5fc74c3 mm: hugetlb: Copy... |
4935 |
{ |
e66f17ff7 mm/hugetlb: take ... |
4936 4937 |
if (flags & FOLL_GET) return NULL; |
9e5fc74c3 mm: hugetlb: Copy... |
4938 |
|
e66f17ff7 mm/hugetlb: take ... |
4939 |
return pte_page(*(pte_t *)pud) + ((address & ~PUD_MASK) >> PAGE_SHIFT); |
9e5fc74c3 mm: hugetlb: Copy... |
4940 |
} |
faaa5b62d mm/follow_page_ma... |
4941 4942 4943 4944 4945 4946 4947 4948 |
struct page * __weak follow_huge_pgd(struct mm_struct *mm, unsigned long address, pgd_t *pgd, int flags) { if (flags & FOLL_GET) return NULL; return pte_page(*(pte_t *)pgd) + ((address & ~PGDIR_MASK) >> PAGE_SHIFT); } |
31caf665e mm: migrate: make... |
4949 4950 |
bool isolate_huge_page(struct page *page, struct list_head *list) { |
bcc542223 mm: hugetlb: intr... |
4951 |
bool ret = true; |
309381fea mm: dump page whe... |
4952 |
VM_BUG_ON_PAGE(!PageHead(page), page); |
31caf665e mm: migrate: make... |
4953 |
spin_lock(&hugetlb_lock); |
bcc542223 mm: hugetlb: intr... |
4954 4955 4956 4957 4958 |
if (!page_huge_active(page) || !get_page_unless_zero(page)) { ret = false; goto unlock; } clear_page_huge_active(page); |
31caf665e mm: migrate: make... |
4959 |
list_move_tail(&page->lru, list); |
bcc542223 mm: hugetlb: intr... |
4960 |
unlock: |
31caf665e mm: migrate: make... |
4961 |
spin_unlock(&hugetlb_lock); |
bcc542223 mm: hugetlb: intr... |
4962 |
return ret; |
31caf665e mm: migrate: make... |
4963 4964 4965 4966 |
} void putback_active_hugepage(struct page *page) { |
309381fea mm: dump page whe... |
4967 |
VM_BUG_ON_PAGE(!PageHead(page), page); |
31caf665e mm: migrate: make... |
4968 |
spin_lock(&hugetlb_lock); |
bcc542223 mm: hugetlb: intr... |
4969 |
set_page_huge_active(page); |
31caf665e mm: migrate: make... |
4970 4971 4972 4973 |
list_move_tail(&page->lru, &(page_hstate(page))->hugepage_activelist); spin_unlock(&hugetlb_lock); put_page(page); } |
ab5ac90ae mm, hugetlb: do n... |
4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 |
void move_hugetlb_state(struct page *oldpage, struct page *newpage, int reason) { struct hstate *h = page_hstate(oldpage); hugetlb_cgroup_migrate(oldpage, newpage); set_page_owner_migrate_reason(newpage, reason); /* * transfer temporary state of the new huge page. This is * reverse to other transitions because the newpage is going to * be final while the old one will be freed so it takes over * the temporary status. * * Also note that we have to transfer the per-node surplus state * here as well otherwise the global surplus count will not match * the per-node's. */ if (PageHugeTemporary(newpage)) { int old_nid = page_to_nid(oldpage); int new_nid = page_to_nid(newpage); SetPageHugeTemporary(oldpage); ClearPageHugeTemporary(newpage); spin_lock(&hugetlb_lock); if (h->surplus_huge_pages_node[old_nid]) { h->surplus_huge_pages_node[old_nid]--; h->surplus_huge_pages_node[new_nid]++; } spin_unlock(&hugetlb_lock); } } |