Blame view
mm/hugetlb.c
123 KB
1da177e4c Linux-2.6.12-rc2 |
1 2 |
/* * Generic hugetlb support. |
6d49e352a propagate name ch... |
3 |
* (C) Nadia Yvette Chambers, April 2004 |
1da177e4c Linux-2.6.12-rc2 |
4 |
*/ |
1da177e4c Linux-2.6.12-rc2 |
5 6 |
#include <linux/list.h> #include <linux/init.h> |
1da177e4c Linux-2.6.12-rc2 |
7 |
#include <linux/mm.h> |
e1759c215 proc: switch /pro... |
8 |
#include <linux/seq_file.h> |
1da177e4c Linux-2.6.12-rc2 |
9 10 |
#include <linux/sysctl.h> #include <linux/highmem.h> |
cddb8a5c1 mmu-notifiers: core |
11 |
#include <linux/mmu_notifier.h> |
1da177e4c Linux-2.6.12-rc2 |
12 |
#include <linux/nodemask.h> |
63551ae0f [PATCH] Hugepage ... |
13 |
#include <linux/pagemap.h> |
5da7ca860 [PATCH] Add NUMA ... |
14 |
#include <linux/mempolicy.h> |
3b32123d7 mm: use macros fr... |
15 |
#include <linux/compiler.h> |
aea47ff36 [PATCH] mm: make ... |
16 |
#include <linux/cpuset.h> |
3935baa9b [PATCH] hugepage:... |
17 |
#include <linux/mutex.h> |
aa888a749 hugetlb: support ... |
18 |
#include <linux/bootmem.h> |
a34378701 hugetlb: new sysf... |
19 |
#include <linux/sysfs.h> |
5a0e3ad6a include cleanup: ... |
20 |
#include <linux/slab.h> |
0fe6e20b9 hugetlb, rmap: ad... |
21 |
#include <linux/rmap.h> |
fd6a03edd HWPOISON, hugetlb... |
22 23 |
#include <linux/swap.h> #include <linux/swapops.h> |
c8721bbbd mm: memory-hotplu... |
24 |
#include <linux/page-isolation.h> |
8382d914e mm, hugetlb: impr... |
25 |
#include <linux/jhash.h> |
d6606683a Revert duplicate ... |
26 |
|
63551ae0f [PATCH] Hugepage ... |
27 28 |
#include <asm/page.h> #include <asm/pgtable.h> |
24669e584 hugetlb: use mmu_... |
29 |
#include <asm/tlb.h> |
63551ae0f [PATCH] Hugepage ... |
30 |
|
24669e584 hugetlb: use mmu_... |
31 |
#include <linux/io.h> |
63551ae0f [PATCH] Hugepage ... |
32 |
#include <linux/hugetlb.h> |
9dd540e23 hugetlb/cgroup: a... |
33 |
#include <linux/hugetlb_cgroup.h> |
9a3052306 hugetlb: add per ... |
34 |
#include <linux/node.h> |
7835e98b2 [PATCH] remove se... |
35 |
#include "internal.h" |
1da177e4c Linux-2.6.12-rc2 |
36 |
|
753162cd8 mm: hugetlb: fix ... |
37 |
int hugepages_treat_as_movable; |
a55164389 hugetlb: modular ... |
38 |
|
c3f38a387 hugetlb: make som... |
39 |
int hugetlb_max_hstate __read_mostly; |
e5ff21594 hugetlb: multiple... |
40 41 |
unsigned int default_hstate_idx; struct hstate hstates[HUGE_MAX_HSTATE]; |
641844f56 mm/hugetlb: intro... |
42 43 44 45 46 |
/* * 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... |
47 |
|
53ba51d21 hugetlb: allow ar... |
48 |
__initdata LIST_HEAD(huge_boot_pages); |
e5ff21594 hugetlb: multiple... |
49 50 51 |
/* for command line parsing */ static struct hstate * __initdata parsed_hstate; static unsigned long __initdata default_hstate_max_huge_pages; |
e11bfbfcb hugetlb: override... |
52 |
static unsigned long __initdata default_hstate_size; |
9fee021d1 mm/hugetlb: intro... |
53 |
static bool __initdata parsed_valid_hugepagesz = true; |
e5ff21594 hugetlb: multiple... |
54 |
|
3935baa9b [PATCH] hugepage:... |
55 |
/* |
31caf665e mm: migrate: make... |
56 57 |
* Protects updates to hugepage_freelists, hugepage_activelist, nr_huge_pages, * free_huge_pages, and surplus_huge_pages. |
3935baa9b [PATCH] hugepage:... |
58 |
*/ |
c3f38a387 hugetlb: make som... |
59 |
DEFINE_SPINLOCK(hugetlb_lock); |
0bd0f9fb1 [PATCH] hugetlb: ... |
60 |
|
8382d914e mm, hugetlb: impr... |
61 62 63 64 65 |
/* * 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... |
66 |
struct mutex *hugetlb_fault_mutex_table ____cacheline_aligned_in_smp; |
8382d914e mm, hugetlb: impr... |
67 |
|
7ca02d0ae hugetlbfs: accept... |
68 69 |
/* Forward declaration */ static int hugetlb_acct_memory(struct hstate *h, long delta); |
90481622d hugepages: fix us... |
70 71 72 73 74 75 76 |
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... |
77 78 79 80 81 82 |
* 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... |
83 |
kfree(spool); |
7ca02d0ae hugetlbfs: accept... |
84 |
} |
90481622d hugepages: fix us... |
85 |
} |
7ca02d0ae hugetlbfs: accept... |
86 87 |
struct hugepage_subpool *hugepage_new_subpool(struct hstate *h, long max_hpages, long min_hpages) |
90481622d hugepages: fix us... |
88 89 |
{ struct hugepage_subpool *spool; |
c6a918200 hugetlbfs: add mi... |
90 |
spool = kzalloc(sizeof(*spool), GFP_KERNEL); |
90481622d hugepages: fix us... |
91 92 93 94 95 |
if (!spool) return NULL; spin_lock_init(&spool->lock); spool->count = 1; |
7ca02d0ae hugetlbfs: accept... |
96 97 98 99 100 101 102 103 104 |
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... |
105 106 107 108 109 110 111 112 113 114 115 |
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... |
116 117 118 119 120 121 122 123 124 |
/* * 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... |
125 126 |
long delta) { |
1c5ecae3a hugetlbfs: add mi... |
127 |
long ret = delta; |
90481622d hugepages: fix us... |
128 129 |
if (!spool) |
1c5ecae3a hugetlbfs: add mi... |
130 |
return ret; |
90481622d hugepages: fix us... |
131 132 |
spin_lock(&spool->lock); |
1c5ecae3a hugetlbfs: add mi... |
133 134 135 136 137 138 139 140 |
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... |
141 |
} |
90481622d hugepages: fix us... |
142 |
|
09a95e29c mm/hugetlb: optim... |
143 144 |
/* minimum size accounting */ if (spool->min_hpages != -1 && spool->rsv_hpages) { |
1c5ecae3a hugetlbfs: add mi... |
145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 |
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... |
160 161 |
return ret; } |
1c5ecae3a hugetlbfs: add mi... |
162 163 164 165 166 167 168 |
/* * 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... |
169 170 |
long delta) { |
1c5ecae3a hugetlbfs: add mi... |
171 |
long ret = delta; |
90481622d hugepages: fix us... |
172 |
if (!spool) |
1c5ecae3a hugetlbfs: add mi... |
173 |
return delta; |
90481622d hugepages: fix us... |
174 175 |
spin_lock(&spool->lock); |
1c5ecae3a hugetlbfs: add mi... |
176 177 178 |
if (spool->max_hpages != -1) /* maximum size accounting */ spool->used_hpages -= delta; |
09a95e29c mm/hugetlb: optim... |
179 180 |
/* minimum size accounting */ if (spool->min_hpages != -1 && spool->used_hpages < spool->min_hpages) { |
1c5ecae3a hugetlbfs: add mi... |
181 182 183 184 185 186 187 188 189 190 191 192 193 194 |
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... |
195 |
unlock_or_release_subpool(spool); |
1c5ecae3a hugetlbfs: add mi... |
196 197 |
return ret; |
90481622d hugepages: fix us... |
198 199 200 201 202 203 204 205 206 |
} 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_... |
207 |
return subpool_inode(file_inode(vma->vm_file)); |
90481622d hugepages: fix us... |
208 |
} |
e7c4b0bfd huge page private... |
209 |
/* |
968229048 hugetlb: move res... |
210 211 |
* Region tracking -- allows tracking of reservations and instantiated pages * across the pages in a mapping. |
84afd99b8 hugetlb reservati... |
212 |
* |
1dd308a7b mm/hugetlb: docum... |
213 214 215 216 217 218 219 220 221 222 223 224 225 226 |
* 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... |
227 228 229 230 231 232 |
*/ struct file_region { struct list_head link; long from; long to; }; |
1dd308a7b mm/hugetlb: docum... |
233 234 |
/* * Add the huge page range represented by [f, t) to the reserve |
5e9113731 mm/hugetlb: add c... |
235 236 237 238 239 240 241 242 |
* 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... |
243 244 245 |
* * Return the number of new huge pages added to the map. This * number is greater than or equal to zero. |
1dd308a7b mm/hugetlb: docum... |
246 |
*/ |
1406ec9ba mm, hugetlb: impr... |
247 |
static long region_add(struct resv_map *resv, long f, long t) |
968229048 hugetlb: move res... |
248 |
{ |
1406ec9ba mm, hugetlb: impr... |
249 |
struct list_head *head = &resv->regions; |
968229048 hugetlb: move res... |
250 |
struct file_region *rg, *nrg, *trg; |
cf3ad20bf mm/hugetlb: compu... |
251 |
long add = 0; |
968229048 hugetlb: move res... |
252 |
|
7b24d8616 mm, hugetlb: fix ... |
253 |
spin_lock(&resv->lock); |
968229048 hugetlb: move res... |
254 255 256 257 |
/* 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... |
258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 |
/* * 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... |
279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 |
/* 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... |
297 298 299 300 301 |
/* 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... |
302 303 304 305 |
list_del(&rg->link); kfree(rg); } } |
cf3ad20bf mm/hugetlb: compu... |
306 307 |
add += (nrg->from - f); /* Added to beginning of region */ |
968229048 hugetlb: move res... |
308 |
nrg->from = f; |
cf3ad20bf mm/hugetlb: compu... |
309 |
add += t - nrg->to; /* Added to end of region */ |
968229048 hugetlb: move res... |
310 |
nrg->to = t; |
cf3ad20bf mm/hugetlb: compu... |
311 |
|
5e9113731 mm/hugetlb: add c... |
312 313 |
out_locked: resv->adds_in_progress--; |
7b24d8616 mm, hugetlb: fix ... |
314 |
spin_unlock(&resv->lock); |
cf3ad20bf mm/hugetlb: compu... |
315 316 |
VM_BUG_ON(add < 0); return add; |
968229048 hugetlb: move res... |
317 |
} |
1dd308a7b mm/hugetlb: docum... |
318 319 320 321 322 323 324 325 326 327 328 329 330 |
/* * 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... |
331 332 333 334 335 336 337 338 |
* 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... |
339 |
*/ |
1406ec9ba mm, hugetlb: impr... |
340 |
static long region_chg(struct resv_map *resv, long f, long t) |
968229048 hugetlb: move res... |
341 |
{ |
1406ec9ba mm, hugetlb: impr... |
342 |
struct list_head *head = &resv->regions; |
7b24d8616 mm, hugetlb: fix ... |
343 |
struct file_region *rg, *nrg = NULL; |
968229048 hugetlb: move res... |
344 |
long chg = 0; |
7b24d8616 mm, hugetlb: fix ... |
345 346 |
retry: spin_lock(&resv->lock); |
5e9113731 mm/hugetlb: add c... |
347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 |
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... |
363 364 |
if (!trg) { kfree(nrg); |
5e9113731 mm/hugetlb: add c... |
365 |
return -ENOMEM; |
dbe409e4f mm/hugetlb.c: fix... |
366 |
} |
5e9113731 mm/hugetlb: add c... |
367 368 369 370 371 372 |
spin_lock(&resv->lock); list_add(&trg->link, &resv->region_cache); resv->region_cache_count++; goto retry_locked; } |
968229048 hugetlb: move res... |
373 374 375 376 377 378 379 380 381 |
/* 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 ... |
382 |
if (!nrg) { |
5e9113731 mm/hugetlb: add c... |
383 |
resv->adds_in_progress--; |
7b24d8616 mm, hugetlb: fix ... |
384 385 386 387 388 389 390 391 392 393 |
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... |
394 |
|
7b24d8616 mm, hugetlb: fix ... |
395 396 397 |
list_add(&nrg->link, rg->link.prev); chg = t - f; goto out_nrg; |
968229048 hugetlb: move res... |
398 399 400 401 402 403 404 405 406 407 408 409 |
} /* 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 ... |
410 |
goto out; |
968229048 hugetlb: move res... |
411 |
|
25985edce Fix common misspe... |
412 |
/* We overlap with this area, if it extends further than |
968229048 hugetlb: move res... |
413 414 415 416 417 418 419 420 |
* 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 ... |
421 422 423 424 425 426 427 428 |
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... |
429 430 |
return chg; } |
1dd308a7b mm/hugetlb: docum... |
431 |
/* |
5e9113731 mm/hugetlb: add c... |
432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 |
* 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... |
451 452 453 454 455 456 457 458 459 460 461 462 |
* 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... |
463 |
*/ |
feba16e25 mm/hugetlb: add r... |
464 |
static long region_del(struct resv_map *resv, long f, long t) |
968229048 hugetlb: move res... |
465 |
{ |
1406ec9ba mm, hugetlb: impr... |
466 |
struct list_head *head = &resv->regions; |
968229048 hugetlb: move res... |
467 |
struct file_region *rg, *trg; |
feba16e25 mm/hugetlb: add r... |
468 469 |
struct file_region *nrg = NULL; long del = 0; |
968229048 hugetlb: move res... |
470 |
|
feba16e25 mm/hugetlb: add r... |
471 |
retry: |
7b24d8616 mm, hugetlb: fix ... |
472 |
spin_lock(&resv->lock); |
feba16e25 mm/hugetlb: add r... |
473 |
list_for_each_entry_safe(rg, trg, head, link) { |
dbe409e4f mm/hugetlb.c: fix... |
474 475 476 477 478 479 480 481 |
/* * 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... |
482 |
continue; |
dbe409e4f mm/hugetlb.c: fix... |
483 |
|
feba16e25 mm/hugetlb: add r... |
484 |
if (rg->from >= t) |
968229048 hugetlb: move res... |
485 |
break; |
968229048 hugetlb: move res... |
486 |
|
feba16e25 mm/hugetlb: add r... |
487 488 489 490 491 492 493 494 495 496 497 498 499 |
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... |
500 |
|
feba16e25 mm/hugetlb: add r... |
501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 |
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... |
521 |
break; |
feba16e25 mm/hugetlb: add r... |
522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 |
} 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... |
538 |
} |
7b24d8616 mm, hugetlb: fix ... |
539 |
|
7b24d8616 mm, hugetlb: fix ... |
540 |
spin_unlock(&resv->lock); |
feba16e25 mm/hugetlb: add r... |
541 542 |
kfree(nrg); return del; |
968229048 hugetlb: move res... |
543 |
} |
1dd308a7b mm/hugetlb: docum... |
544 |
/* |
b5cec28d3 hugetlbfs: trunca... |
545 546 547 548 549 550 551 552 |
* 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... |
553 |
void hugetlb_fix_reserve_counts(struct inode *inode) |
b5cec28d3 hugetlbfs: trunca... |
554 555 556 557 558 |
{ struct hugepage_subpool *spool = subpool_inode(inode); long rsv_adjust; rsv_adjust = hugepage_subpool_get_pages(spool, 1); |
72e2936c0 mm: remove unnece... |
559 |
if (rsv_adjust) { |
b5cec28d3 hugetlbfs: trunca... |
560 561 562 563 564 565 566 |
struct hstate *h = hstate_inode(inode); hugetlb_acct_memory(h, 1); } } /* |
1dd308a7b mm/hugetlb: docum... |
567 568 569 |
* Count and return the number of huge pages in the reserve map * that intersect with the range [f, t). */ |
1406ec9ba mm, hugetlb: impr... |
570 |
static long region_count(struct resv_map *resv, long f, long t) |
84afd99b8 hugetlb reservati... |
571 |
{ |
1406ec9ba mm, hugetlb: impr... |
572 |
struct list_head *head = &resv->regions; |
84afd99b8 hugetlb reservati... |
573 574 |
struct file_region *rg; long chg = 0; |
7b24d8616 mm, hugetlb: fix ... |
575 |
spin_lock(&resv->lock); |
84afd99b8 hugetlb reservati... |
576 577 |
/* Locate each segment we overlap with, and count that overlap. */ list_for_each_entry(rg, head, link) { |
f2135a4a5 mm/hugetlb.c: use... |
578 579 |
long seg_from; long seg_to; |
84afd99b8 hugetlb reservati... |
580 581 582 583 584 585 586 587 588 589 590 |
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 ... |
591 |
spin_unlock(&resv->lock); |
84afd99b8 hugetlb reservati... |
592 593 594 |
return chg; } |
968229048 hugetlb: move res... |
595 |
/* |
e7c4b0bfd huge page private... |
596 |
* Convert the address within this vma to the page offset within |
e7c4b0bfd huge page private... |
597 598 |
* the mapping, in pagecache page units; huge pages here. */ |
a55164389 hugetlb: modular ... |
599 600 |
static pgoff_t vma_hugecache_offset(struct hstate *h, struct vm_area_struct *vma, unsigned long address) |
e7c4b0bfd huge page private... |
601 |
{ |
a55164389 hugetlb: modular ... |
602 603 |
return ((address - vma->vm_start) >> huge_page_shift(h)) + (vma->vm_pgoff >> huge_page_order(h)); |
e7c4b0bfd huge page private... |
604 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
605 606 607 608 609 |
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... |
610 |
EXPORT_SYMBOL_GPL(linear_hugepage_index); |
0fe6e20b9 hugetlb, rmap: ad... |
611 |
|
84afd99b8 hugetlb reservati... |
612 |
/* |
08fba6998 mm: report the pa... |
613 614 615 616 617 618 619 620 621 622 623 |
* Return the size of the pages allocated when backing a VMA. In the majority * cases this will be same size as used by the page table entries. */ unsigned long vma_kernel_pagesize(struct vm_area_struct *vma) { struct hstate *hstate; if (!is_vm_hugetlb_page(vma)) return PAGE_SIZE; hstate = hstate_vma(vma); |
2415cf12e mm/hugetlb: use a... |
624 |
return 1UL << huge_page_shift(hstate); |
08fba6998 mm: report the pa... |
625 |
} |
f340ca0f0 hugetlbfs: export... |
626 |
EXPORT_SYMBOL_GPL(vma_kernel_pagesize); |
08fba6998 mm: report the pa... |
627 628 |
/* |
3340289dd mm: report the MM... |
629 630 631 632 633 634 635 636 637 638 639 640 641 |
* Return the page size being used by the MMU to back a VMA. In the majority * of cases, the page size used by the kernel matches the MMU size. On * architectures where it differs, an architecture-specific version of this * function is required. */ #ifndef vma_mmu_pagesize unsigned long vma_mmu_pagesize(struct vm_area_struct *vma) { return vma_kernel_pagesize(vma); } #endif /* |
84afd99b8 hugetlb reservati... |
642 643 644 645 646 647 |
* 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... |
648 |
#define HPAGE_RESV_MASK (HPAGE_RESV_OWNER | HPAGE_RESV_UNMAPPED) |
84afd99b8 hugetlb reservati... |
649 |
|
a1e78772d hugetlb: reserve ... |
650 651 652 653 654 655 656 657 658 |
/* * 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... |
659 660 661 662 663 664 665 666 667 |
* * 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 ... |
668 |
*/ |
e7c4b0bfd huge page private... |
669 670 671 672 673 674 675 676 677 678 |
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... |
679 |
struct resv_map *resv_map_alloc(void) |
84afd99b8 hugetlb reservati... |
680 681 |
{ struct resv_map *resv_map = kmalloc(sizeof(*resv_map), GFP_KERNEL); |
5e9113731 mm/hugetlb: add c... |
682 683 684 685 686 |
struct file_region *rg = kmalloc(sizeof(*rg), GFP_KERNEL); if (!resv_map || !rg) { kfree(resv_map); kfree(rg); |
84afd99b8 hugetlb reservati... |
687 |
return NULL; |
5e9113731 mm/hugetlb: add c... |
688 |
} |
84afd99b8 hugetlb reservati... |
689 690 |
kref_init(&resv_map->refs); |
7b24d8616 mm, hugetlb: fix ... |
691 |
spin_lock_init(&resv_map->lock); |
84afd99b8 hugetlb reservati... |
692 |
INIT_LIST_HEAD(&resv_map->regions); |
5e9113731 mm/hugetlb: add c... |
693 694 695 696 697 |
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... |
698 699 |
return resv_map; } |
9119a41e9 mm, hugetlb: unif... |
700 |
void resv_map_release(struct kref *ref) |
84afd99b8 hugetlb reservati... |
701 702 |
{ struct resv_map *resv_map = container_of(ref, struct resv_map, refs); |
5e9113731 mm/hugetlb: add c... |
703 704 |
struct list_head *head = &resv_map->region_cache; struct file_region *rg, *trg; |
84afd99b8 hugetlb reservati... |
705 706 |
/* Clear out any active regions before we release the map. */ |
feba16e25 mm/hugetlb: add r... |
707 |
region_del(resv_map, 0, LONG_MAX); |
5e9113731 mm/hugetlb: add c... |
708 709 710 711 712 713 714 715 |
/* ... 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... |
716 717 |
kfree(resv_map); } |
4e35f4838 mm, hugetlb: use ... |
718 719 720 721 |
static inline struct resv_map *inode_resv_map(struct inode *inode) { return inode->i_mapping->private_data; } |
84afd99b8 hugetlb reservati... |
722 |
static struct resv_map *vma_resv_map(struct vm_area_struct *vma) |
a1e78772d hugetlb: reserve ... |
723 |
{ |
81d1b09c6 mm: convert a few... |
724 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); |
4e35f4838 mm, hugetlb: use ... |
725 726 727 728 729 730 731 |
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... |
732 733 |
return (struct resv_map *)(get_vma_private_data(vma) & ~HPAGE_RESV_MASK); |
4e35f4838 mm, hugetlb: use ... |
734 |
} |
a1e78772d hugetlb: reserve ... |
735 |
} |
84afd99b8 hugetlb reservati... |
736 |
static void set_vma_resv_map(struct vm_area_struct *vma, struct resv_map *map) |
a1e78772d hugetlb: reserve ... |
737 |
{ |
81d1b09c6 mm: convert a few... |
738 739 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); VM_BUG_ON_VMA(vma->vm_flags & VM_MAYSHARE, vma); |
a1e78772d hugetlb: reserve ... |
740 |
|
84afd99b8 hugetlb reservati... |
741 742 |
set_vma_private_data(vma, (get_vma_private_data(vma) & HPAGE_RESV_MASK) | (unsigned long)map); |
04f2cbe35 hugetlb: guarante... |
743 744 745 746 |
} static void set_vma_resv_flags(struct vm_area_struct *vma, unsigned long flags) { |
81d1b09c6 mm: convert a few... |
747 748 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); VM_BUG_ON_VMA(vma->vm_flags & VM_MAYSHARE, vma); |
e7c4b0bfd huge page private... |
749 750 |
set_vma_private_data(vma, get_vma_private_data(vma) | flags); |
04f2cbe35 hugetlb: guarante... |
751 752 753 754 |
} static int is_vma_resv_set(struct vm_area_struct *vma, unsigned long flag) { |
81d1b09c6 mm: convert a few... |
755 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); |
e7c4b0bfd huge page private... |
756 757 |
return (get_vma_private_data(vma) & flag) != 0; |
a1e78772d hugetlb: reserve ... |
758 |
} |
04f2cbe35 hugetlb: guarante... |
759 |
/* Reset counters to 0 and clear all HPAGE_RESV_* flags */ |
a1e78772d hugetlb: reserve ... |
760 761 |
void reset_vma_resv_huge_pages(struct vm_area_struct *vma) { |
81d1b09c6 mm: convert a few... |
762 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); |
f83a275db mm: account for M... |
763 |
if (!(vma->vm_flags & VM_MAYSHARE)) |
a1e78772d hugetlb: reserve ... |
764 765 766 767 |
vma->vm_private_data = (void *)0; } /* Returns true if the VMA has associated reserve pages */ |
559ec2f8f mm/hugetlb.c: mak... |
768 |
static bool vma_has_reserves(struct vm_area_struct *vma, long chg) |
a1e78772d hugetlb: reserve ... |
769 |
{ |
af0ed73e6 mm, hugetlb: decr... |
770 771 772 773 774 775 776 777 778 779 780 |
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... |
781 |
return true; |
af0ed73e6 mm, hugetlb: decr... |
782 |
else |
559ec2f8f mm/hugetlb.c: mak... |
783 |
return false; |
af0ed73e6 mm, hugetlb: decr... |
784 |
} |
a63884e92 mm, hugetlb: remo... |
785 786 |
/* Shared mappings always use reserves */ |
1fb1b0e9e mm/hugetlb: vma_h... |
787 788 789 790 791 792 793 794 795 796 797 798 799 |
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... |
800 801 802 803 804 |
/* * Only the process that called mmap() has reserves for * private mappings. */ |
67961f9db mm/hugetlb: fix h... |
805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 |
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... |
826 |
|
559ec2f8f mm/hugetlb.c: mak... |
827 |
return false; |
a1e78772d hugetlb: reserve ... |
828 |
} |
a55164389 hugetlb: modular ... |
829 |
static void enqueue_huge_page(struct hstate *h, struct page *page) |
1da177e4c Linux-2.6.12-rc2 |
830 831 |
{ int nid = page_to_nid(page); |
0edaecfab hugetlb: add a li... |
832 |
list_move(&page->lru, &h->hugepage_freelists[nid]); |
a55164389 hugetlb: modular ... |
833 834 |
h->free_huge_pages++; h->free_huge_pages_node[nid]++; |
1da177e4c Linux-2.6.12-rc2 |
835 |
} |
bf50bab2b hugetlb: add allo... |
836 837 838 |
static struct page *dequeue_huge_page_node(struct hstate *h, int nid) { struct page *page; |
c8721bbbd mm: memory-hotplu... |
839 840 841 842 843 844 845 846 |
list_for_each_entry(page, &h->hugepage_freelists[nid], lru) if (!is_migrate_isolate_page(page)) break; /* * if 'non-isolated free hugepage' not found on the list, * the allocation fails. */ if (&h->hugepage_freelists[nid] == &page->lru) |
bf50bab2b hugetlb: add allo... |
847 |
return NULL; |
0edaecfab hugetlb: add a li... |
848 |
list_move(&page->lru, &h->hugepage_activelist); |
a9869b837 hugetlb: move ref... |
849 |
set_page_refcounted(page); |
bf50bab2b hugetlb: add allo... |
850 851 852 853 |
h->free_huge_pages--; h->free_huge_pages_node[nid]--; return page; } |
86cdb465c mm: prepare to re... |
854 855 856 |
/* Movability of hugepages depends on migration support. */ static inline gfp_t htlb_alloc_mask(struct hstate *h) { |
100873d7a hugetlb: rename h... |
857 |
if (hugepages_treat_as_movable || hugepage_migration_supported(h)) |
86cdb465c mm: prepare to re... |
858 859 860 861 |
return GFP_HIGHUSER_MOVABLE; else return GFP_HIGHUSER; } |
a55164389 hugetlb: modular ... |
862 863 |
static struct page *dequeue_huge_page_vma(struct hstate *h, struct vm_area_struct *vma, |
af0ed73e6 mm, hugetlb: decr... |
864 865 |
unsigned long address, int avoid_reserve, long chg) |
1da177e4c Linux-2.6.12-rc2 |
866 |
{ |
b1c12cbcd mm/hugetlb: fix w... |
867 |
struct page *page = NULL; |
480eccf9a Fix NUMA Memory P... |
868 |
struct mempolicy *mpol; |
19770b326 mm: filter based ... |
869 |
nodemask_t *nodemask; |
c0ff7453b cpuset,mm: fix no... |
870 |
struct zonelist *zonelist; |
dd1a239f6 mm: have zonelist... |
871 872 |
struct zone *zone; struct zoneref *z; |
cc9a6c877 cpuset: mm: reduc... |
873 |
unsigned int cpuset_mems_cookie; |
1da177e4c Linux-2.6.12-rc2 |
874 |
|
a1e78772d hugetlb: reserve ... |
875 876 877 878 879 |
/* * 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... |
880 |
if (!vma_has_reserves(vma, chg) && |
a55164389 hugetlb: modular ... |
881 |
h->free_huge_pages - h->resv_huge_pages == 0) |
c0ff7453b cpuset,mm: fix no... |
882 |
goto err; |
a1e78772d hugetlb: reserve ... |
883 |
|
04f2cbe35 hugetlb: guarante... |
884 |
/* If reserves cannot be used, ensure enough pages are in the pool */ |
a55164389 hugetlb: modular ... |
885 |
if (avoid_reserve && h->free_huge_pages - h->resv_huge_pages == 0) |
6eab04a87 treewide: remove ... |
886 |
goto err; |
04f2cbe35 hugetlb: guarante... |
887 |
|
9966c4bbb mm, hugetlb: move... |
888 |
retry_cpuset: |
d26914d11 mm: optimize put_... |
889 |
cpuset_mems_cookie = read_mems_allowed_begin(); |
9966c4bbb mm, hugetlb: move... |
890 |
zonelist = huge_zonelist(vma, address, |
86cdb465c mm: prepare to re... |
891 |
htlb_alloc_mask(h), &mpol, &nodemask); |
9966c4bbb mm, hugetlb: move... |
892 |
|
19770b326 mm: filter based ... |
893 894 |
for_each_zone_zonelist_nodemask(zone, z, zonelist, MAX_NR_ZONES - 1, nodemask) { |
344736f29 cpuset: simplify ... |
895 |
if (cpuset_zone_allowed(zone, htlb_alloc_mask(h))) { |
bf50bab2b hugetlb: add allo... |
896 897 |
page = dequeue_huge_page_node(h, zone_to_nid(zone)); if (page) { |
af0ed73e6 mm, hugetlb: decr... |
898 899 900 901 |
if (avoid_reserve) break; if (!vma_has_reserves(vma, chg)) break; |
07443a85a mm, hugetlb: retu... |
902 |
SetPagePrivate(page); |
af0ed73e6 mm, hugetlb: decr... |
903 |
h->resv_huge_pages--; |
bf50bab2b hugetlb: add allo... |
904 905 |
break; } |
3abf7afd4 dequeue_huge_page... |
906 |
} |
1da177e4c Linux-2.6.12-rc2 |
907 |
} |
cc9a6c877 cpuset: mm: reduc... |
908 |
|
52cd3b074 mempolicy: rework... |
909 |
mpol_cond_put(mpol); |
d26914d11 mm: optimize put_... |
910 |
if (unlikely(!page && read_mems_allowed_retry(cpuset_mems_cookie))) |
cc9a6c877 cpuset: mm: reduc... |
911 |
goto retry_cpuset; |
1da177e4c Linux-2.6.12-rc2 |
912 |
return page; |
cc9a6c877 cpuset: mm: reduc... |
913 914 |
err: |
cc9a6c877 cpuset: mm: reduc... |
915 |
return NULL; |
1da177e4c Linux-2.6.12-rc2 |
916 |
} |
1cac6f2c0 hugetlb: move hel... |
917 918 919 920 921 922 923 924 925 |
/* * 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... |
926 |
nid = next_node_in(nid, *nodes_allowed); |
1cac6f2c0 hugetlb: move hel... |
927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 |
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--) |
461a71843 mm/hugetlb: intro... |
987 |
#if defined(CONFIG_ARCH_HAS_GIGANTIC_PAGE) && \ |
d08de8e2d s390/mm: add supp... |
988 989 |
((defined(CONFIG_MEMORY_ISOLATION) && defined(CONFIG_COMPACTION)) || \ defined(CONFIG_CMA)) |
944d9fec8 hugetlb: add supp... |
990 |
static void destroy_compound_gigantic_page(struct page *page, |
d00181b96 mm: use 'unsigned... |
991 |
unsigned int order) |
944d9fec8 hugetlb: add supp... |
992 993 994 995 |
{ int i; int nr_pages = 1 << order; struct page *p = page + 1; |
c8cc708a3 mm/hugetlb: clear... |
996 |
atomic_set(compound_mapcount_ptr(page), 0); |
944d9fec8 hugetlb: add supp... |
997 |
for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) { |
1d798ca3f mm: make compound... |
998 |
clear_compound_head(p); |
944d9fec8 hugetlb: add supp... |
999 |
set_page_refcounted(p); |
944d9fec8 hugetlb: add supp... |
1000 1001 1002 1003 1004 |
} set_compound_order(page, 0); __ClearPageHead(page); } |
d00181b96 mm: use 'unsigned... |
1005 |
static void free_gigantic_page(struct page *page, unsigned int order) |
944d9fec8 hugetlb: add supp... |
1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 |
{ free_contig_range(page_to_pfn(page), 1 << order); } static int __alloc_gigantic_page(unsigned long start_pfn, unsigned long nr_pages) { unsigned long end_pfn = start_pfn + nr_pages; return alloc_contig_range(start_pfn, end_pfn, MIGRATE_MOVABLE); } |
f44b2dda8 mm/hugetlb: add s... |
1016 1017 |
static bool pfn_range_valid_gigantic(struct zone *z, unsigned long start_pfn, unsigned long nr_pages) |
944d9fec8 hugetlb: add supp... |
1018 1019 1020 1021 1022 1023 1024 1025 1026 |
{ unsigned long i, end_pfn = start_pfn + nr_pages; struct page *page; for (i = start_pfn; i < end_pfn; i++) { if (!pfn_valid(i)) return false; page = pfn_to_page(i); |
f44b2dda8 mm/hugetlb: add s... |
1027 1028 |
if (page_zone(page) != z) return false; |
944d9fec8 hugetlb: add supp... |
1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 |
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); } |
d00181b96 mm: use 'unsigned... |
1048 |
static struct page *alloc_gigantic_page(int nid, unsigned int order) |
944d9fec8 hugetlb: add supp... |
1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 |
{ unsigned long nr_pages = 1 << order; unsigned long ret, pfn, flags; struct zone *z; z = NODE_DATA(nid)->node_zones; for (; z - NODE_DATA(nid)->node_zones < MAX_NR_ZONES; z++) { spin_lock_irqsave(&z->lock, flags); pfn = ALIGN(z->zone_start_pfn, nr_pages); while (zone_spans_last_pfn(z, pfn, nr_pages)) { |
f44b2dda8 mm/hugetlb: add s... |
1060 |
if (pfn_range_valid_gigantic(z, pfn, nr_pages)) { |
944d9fec8 hugetlb: add supp... |
1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 |
/* * We release the zone lock here because * alloc_contig_range() will also lock the zone * at some point. If there's an allocation * spinning on this lock, it may win the race * and cause alloc_contig_range() to fail... */ spin_unlock_irqrestore(&z->lock, flags); ret = __alloc_gigantic_page(pfn, nr_pages); if (!ret) return pfn_to_page(pfn); spin_lock_irqsave(&z->lock, flags); } pfn += nr_pages; } spin_unlock_irqrestore(&z->lock, flags); } return NULL; } static void prep_new_huge_page(struct hstate *h, struct page *page, int nid); |
d00181b96 mm: use 'unsigned... |
1084 |
static void prep_compound_gigantic_page(struct page *page, unsigned int order); |
944d9fec8 hugetlb: add supp... |
1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 |
static struct page *alloc_fresh_gigantic_page_node(struct hstate *h, int nid) { struct page *page; page = alloc_gigantic_page(nid, huge_page_order(h)); if (page) { prep_compound_gigantic_page(page, huge_page_order(h)); prep_new_huge_page(h, page, nid); } return page; } static int alloc_fresh_gigantic_page(struct hstate *h, nodemask_t *nodes_allowed) { struct page *page = NULL; int nr_nodes, node; for_each_node_mask_to_alloc(h, nr_nodes, node, nodes_allowed) { page = alloc_fresh_gigantic_page_node(h, node); if (page) return 1; } return 0; } static inline bool gigantic_page_supported(void) { return true; } #else static inline bool gigantic_page_supported(void) { return false; } |
d00181b96 mm: use 'unsigned... |
1117 |
static inline void free_gigantic_page(struct page *page, unsigned int order) { } |
944d9fec8 hugetlb: add supp... |
1118 |
static inline void destroy_compound_gigantic_page(struct page *page, |
d00181b96 mm: use 'unsigned... |
1119 |
unsigned int order) { } |
944d9fec8 hugetlb: add supp... |
1120 1121 1122 |
static inline int alloc_fresh_gigantic_page(struct hstate *h, nodemask_t *nodes_allowed) { return 0; } #endif |
a55164389 hugetlb: modular ... |
1123 |
static void update_and_free_page(struct hstate *h, struct page *page) |
6af2acb66 hugetlb: Move upd... |
1124 1125 |
{ int i; |
a55164389 hugetlb: modular ... |
1126 |
|
944d9fec8 hugetlb: add supp... |
1127 1128 |
if (hstate_is_gigantic(h) && !gigantic_page_supported()) return; |
18229df5b hugetlb: pull gig... |
1129 |
|
a55164389 hugetlb: modular ... |
1130 1131 1132 |
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 ... |
1133 1134 |
page[i].flags &= ~(1 << PG_locked | 1 << PG_error | 1 << PG_referenced | 1 << PG_dirty | |
a7407a27c hugetlb: update_a... |
1135 1136 |
1 << PG_active | 1 << PG_private | 1 << PG_writeback); |
6af2acb66 hugetlb: Move upd... |
1137 |
} |
309381fea mm: dump page whe... |
1138 |
VM_BUG_ON_PAGE(hugetlb_cgroup_from_page(page), page); |
f1e61557f mm: pack compound... |
1139 |
set_compound_page_dtor(page, NULL_COMPOUND_DTOR); |
6af2acb66 hugetlb: Move upd... |
1140 |
set_page_refcounted(page); |
944d9fec8 hugetlb: add supp... |
1141 1142 1143 1144 |
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... |
1145 1146 |
__free_pages(page, huge_page_order(h)); } |
6af2acb66 hugetlb: Move upd... |
1147 |
} |
e5ff21594 hugetlb: multiple... |
1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 |
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... |
1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 |
/* * 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]); } |
8f1d26d0e kexec: export fre... |
1182 |
void free_huge_page(struct page *page) |
27a85ef1b [PATCH] hugepage:... |
1183 |
{ |
a55164389 hugetlb: modular ... |
1184 1185 1186 1187 |
/* * Can't pass hstate in here because it is called from the * compound page destructor. */ |
e5ff21594 hugetlb: multiple... |
1188 |
struct hstate *h = page_hstate(page); |
7893d1d50 hugetlb: Try to g... |
1189 |
int nid = page_to_nid(page); |
90481622d hugepages: fix us... |
1190 1191 |
struct hugepage_subpool *spool = (struct hugepage_subpool *)page_private(page); |
07443a85a mm, hugetlb: retu... |
1192 |
bool restore_reserve; |
27a85ef1b [PATCH] hugepage:... |
1193 |
|
e5df70ab1 hugetlb: ensure w... |
1194 |
set_page_private(page, 0); |
23be7468e hugetlb: fix infi... |
1195 |
page->mapping = NULL; |
b4330afbe mm/hugetlb: fix g... |
1196 1197 |
VM_BUG_ON_PAGE(page_count(page), page); VM_BUG_ON_PAGE(page_mapcount(page), page); |
07443a85a mm, hugetlb: retu... |
1198 |
restore_reserve = PagePrivate(page); |
16c794b4f mm/hugetlb.c: cor... |
1199 |
ClearPagePrivate(page); |
27a85ef1b [PATCH] hugepage:... |
1200 |
|
1c5ecae3a hugetlbfs: add mi... |
1201 1202 1203 1204 1205 1206 1207 |
/* * 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; |
27a85ef1b [PATCH] hugepage:... |
1208 |
spin_lock(&hugetlb_lock); |
bcc542223 mm: hugetlb: intr... |
1209 |
clear_page_huge_active(page); |
6d76dcf40 hugetlb/cgroup: a... |
1210 1211 |
hugetlb_cgroup_uncharge_page(hstate_index(h), pages_per_huge_page(h), page); |
07443a85a mm, hugetlb: retu... |
1212 1213 |
if (restore_reserve) h->resv_huge_pages++; |
944d9fec8 hugetlb: add supp... |
1214 |
if (h->surplus_huge_pages_node[nid]) { |
0edaecfab hugetlb: add a li... |
1215 1216 |
/* remove the page from active list */ list_del(&page->lru); |
a55164389 hugetlb: modular ... |
1217 1218 1219 |
update_and_free_page(h, page); h->surplus_huge_pages--; h->surplus_huge_pages_node[nid]--; |
7893d1d50 hugetlb: Try to g... |
1220 |
} else { |
5d3a551c2 mm: hugetlb: add ... |
1221 |
arch_clear_hugepage_flags(page); |
a55164389 hugetlb: modular ... |
1222 |
enqueue_huge_page(h, page); |
7893d1d50 hugetlb: Try to g... |
1223 |
} |
27a85ef1b [PATCH] hugepage:... |
1224 1225 |
spin_unlock(&hugetlb_lock); } |
a55164389 hugetlb: modular ... |
1226 |
static void prep_new_huge_page(struct hstate *h, struct page *page, int nid) |
b7ba30c67 hugetlb: factor o... |
1227 |
{ |
0edaecfab hugetlb: add a li... |
1228 |
INIT_LIST_HEAD(&page->lru); |
f1e61557f mm: pack compound... |
1229 |
set_compound_page_dtor(page, HUGETLB_PAGE_DTOR); |
b7ba30c67 hugetlb: factor o... |
1230 |
spin_lock(&hugetlb_lock); |
9dd540e23 hugetlb/cgroup: a... |
1231 |
set_hugetlb_cgroup(page, NULL); |
a55164389 hugetlb: modular ... |
1232 1233 |
h->nr_huge_pages++; h->nr_huge_pages_node[nid]++; |
b7ba30c67 hugetlb: factor o... |
1234 1235 1236 |
spin_unlock(&hugetlb_lock); put_page(page); /* free it into the hugepage allocator */ } |
d00181b96 mm: use 'unsigned... |
1237 |
static void prep_compound_gigantic_page(struct page *page, unsigned int order) |
20a0307c0 mm: introduce Pag... |
1238 1239 1240 1241 1242 1243 1244 |
{ 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... |
1245 |
__ClearPageReserved(page); |
de09d31dd page-flags: defin... |
1246 |
__SetPageHead(page); |
20a0307c0 mm: introduce Pag... |
1247 |
for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) { |
ef5a22be2 mm: hugetlb: init... |
1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 |
/* * 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... |
1261 |
set_page_count(p, 0); |
1d798ca3f mm: make compound... |
1262 |
set_compound_head(p, page); |
20a0307c0 mm: introduce Pag... |
1263 |
} |
b4330afbe mm/hugetlb: fix g... |
1264 |
atomic_set(compound_mapcount_ptr(page), -1); |
20a0307c0 mm: introduce Pag... |
1265 |
} |
7795912c2 mm: document Page... |
1266 1267 1268 1269 1270 |
/* * 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... |
1271 1272 |
int PageHuge(struct page *page) { |
20a0307c0 mm: introduce Pag... |
1273 1274 1275 1276 |
if (!PageCompound(page)) return 0; page = compound_head(page); |
f1e61557f mm: pack compound... |
1277 |
return page[1].compound_dtor == HUGETLB_PAGE_DTOR; |
20a0307c0 mm: introduce Pag... |
1278 |
} |
43131e141 HWPOISON, hugetlb... |
1279 |
EXPORT_SYMBOL_GPL(PageHuge); |
27c73ae75 mm: hugetlbfs: fi... |
1280 1281 1282 1283 1284 1285 |
/* * 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... |
1286 1287 |
if (!PageHead(page_head)) return 0; |
758f66a29 mm/hugetlb.c: sim... |
1288 |
return get_compound_page_dtor(page_head) == free_huge_page; |
27c73ae75 mm: hugetlbfs: fi... |
1289 |
} |
27c73ae75 mm: hugetlbfs: fi... |
1290 |
|
13d60f4b6 futex: Take hugep... |
1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 |
pgoff_t __basepage_index(struct page *page) { struct page *page_head = compound_head(page); pgoff_t index = page_index(page_head); unsigned long compound_idx; if (!PageHuge(page_head)) return page_index(page); if (compound_order(page_head) >= MAX_ORDER) compound_idx = page_to_pfn(page) - page_to_pfn(page_head); else compound_idx = page - page_head; return (index << compound_order(page_head)) + compound_idx; } |
a55164389 hugetlb: modular ... |
1307 |
static struct page *alloc_fresh_huge_page_node(struct hstate *h, int nid) |
1da177e4c Linux-2.6.12-rc2 |
1308 |
{ |
1da177e4c Linux-2.6.12-rc2 |
1309 |
struct page *page; |
f96efd585 hugetlb: fix race... |
1310 |
|
96db800f5 mm: rename alloc_... |
1311 |
page = __alloc_pages_node(nid, |
86cdb465c mm: prepare to re... |
1312 |
htlb_alloc_mask(h)|__GFP_COMP|__GFP_THISNODE| |
551883ae8 page allocator: e... |
1313 |
__GFP_REPEAT|__GFP_NOWARN, |
a55164389 hugetlb: modular ... |
1314 |
huge_page_order(h)); |
1da177e4c Linux-2.6.12-rc2 |
1315 |
if (page) { |
a55164389 hugetlb: modular ... |
1316 |
prep_new_huge_page(h, page, nid); |
1da177e4c Linux-2.6.12-rc2 |
1317 |
} |
63b4613c3 hugetlb: fix huge... |
1318 1319 1320 |
return page; } |
b22610268 mm, hugetlb: fix ... |
1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 |
static int alloc_fresh_huge_page(struct hstate *h, nodemask_t *nodes_allowed) { struct page *page; int nr_nodes, node; int ret = 0; for_each_node_mask_to_alloc(h, nr_nodes, node, nodes_allowed) { page = alloc_fresh_huge_page_node(h, node); if (page) { ret = 1; break; } } if (ret) count_vm_event(HTLB_BUDDY_PGALLOC); else count_vm_event(HTLB_BUDDY_PGALLOC_FAIL); return ret; } |
e8c5c8249 hugetlb: balance ... |
1342 1343 1344 1345 1346 1347 |
/* * 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... |
1348 1349 |
static int free_pool_huge_page(struct hstate *h, nodemask_t *nodes_allowed, bool acct_surplus) |
e8c5c8249 hugetlb: balance ... |
1350 |
{ |
b22610268 mm, hugetlb: fix ... |
1351 |
int nr_nodes, node; |
e8c5c8249 hugetlb: balance ... |
1352 |
int ret = 0; |
b22610268 mm, hugetlb: fix ... |
1353 |
for_each_node_mask_to_free(h, nr_nodes, node, nodes_allowed) { |
685f34570 hugetlb: use free... |
1354 1355 1356 1357 |
/* * If we're returning unused surplus pages, only examine * nodes with surplus pages. */ |
b22610268 mm, hugetlb: fix ... |
1358 1359 |
if ((!acct_surplus || h->surplus_huge_pages_node[node]) && !list_empty(&h->hugepage_freelists[node])) { |
e8c5c8249 hugetlb: balance ... |
1360 |
struct page *page = |
b22610268 mm, hugetlb: fix ... |
1361 |
list_entry(h->hugepage_freelists[node].next, |
e8c5c8249 hugetlb: balance ... |
1362 1363 1364 |
struct page, lru); list_del(&page->lru); h->free_huge_pages--; |
b22610268 mm, hugetlb: fix ... |
1365 |
h->free_huge_pages_node[node]--; |
685f34570 hugetlb: use free... |
1366 1367 |
if (acct_surplus) { h->surplus_huge_pages--; |
b22610268 mm, hugetlb: fix ... |
1368 |
h->surplus_huge_pages_node[node]--; |
685f34570 hugetlb: use free... |
1369 |
} |
e8c5c8249 hugetlb: balance ... |
1370 1371 |
update_and_free_page(h, page); ret = 1; |
9a76db099 hugetlb: rework h... |
1372 |
break; |
e8c5c8249 hugetlb: balance ... |
1373 |
} |
b22610268 mm, hugetlb: fix ... |
1374 |
} |
e8c5c8249 hugetlb: balance ... |
1375 1376 1377 |
return ret; } |
c8721bbbd mm: memory-hotplu... |
1378 1379 |
/* * Dissolve a given free hugepage into free buddy pages. This function does |
082d5b6b6 mm/hugetlb: check... |
1380 1381 1382 |
* nothing for in-use (including surplus) hugepages. Returns -EBUSY if the * number of free hugepages would be reduced below the number of reserved * hugepages. |
c8721bbbd mm: memory-hotplu... |
1383 |
*/ |
082d5b6b6 mm/hugetlb: check... |
1384 |
static int dissolve_free_huge_page(struct page *page) |
c8721bbbd mm: memory-hotplu... |
1385 |
{ |
082d5b6b6 mm/hugetlb: check... |
1386 |
int rc = 0; |
c8721bbbd mm: memory-hotplu... |
1387 1388 |
spin_lock(&hugetlb_lock); if (PageHuge(page) && !page_count(page)) { |
2247bb335 mm/hugetlb: fix m... |
1389 1390 1391 |
struct page *head = compound_head(page); struct hstate *h = page_hstate(head); int nid = page_to_nid(head); |
082d5b6b6 mm/hugetlb: check... |
1392 1393 1394 1395 |
if (h->free_huge_pages - h->resv_huge_pages == 0) { rc = -EBUSY; goto out; } |
2247bb335 mm/hugetlb: fix m... |
1396 |
list_del(&head->lru); |
c8721bbbd mm: memory-hotplu... |
1397 1398 |
h->free_huge_pages--; h->free_huge_pages_node[nid]--; |
c1470b33b mm/hugetlb: fix i... |
1399 |
h->max_huge_pages--; |
2247bb335 mm/hugetlb: fix m... |
1400 |
update_and_free_page(h, head); |
c8721bbbd mm: memory-hotplu... |
1401 |
} |
082d5b6b6 mm/hugetlb: check... |
1402 |
out: |
c8721bbbd mm: memory-hotplu... |
1403 |
spin_unlock(&hugetlb_lock); |
082d5b6b6 mm/hugetlb: check... |
1404 |
return rc; |
c8721bbbd mm: memory-hotplu... |
1405 1406 1407 1408 1409 |
} /* * 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... |
1410 1411 |
* Note that this will dissolve a free gigantic hugepage completely, if any * part of it lies within the given range. |
082d5b6b6 mm/hugetlb: check... |
1412 1413 |
* 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... |
1414 |
*/ |
082d5b6b6 mm/hugetlb: check... |
1415 |
int dissolve_free_huge_pages(unsigned long start_pfn, unsigned long end_pfn) |
c8721bbbd mm: memory-hotplu... |
1416 |
{ |
c8721bbbd mm: memory-hotplu... |
1417 |
unsigned long pfn; |
eb03aa008 mm/hugetlb: impro... |
1418 |
struct page *page; |
082d5b6b6 mm/hugetlb: check... |
1419 |
int rc = 0; |
c8721bbbd mm: memory-hotplu... |
1420 |
|
d01776393 mm: fix potential... |
1421 |
if (!hugepages_supported()) |
082d5b6b6 mm/hugetlb: check... |
1422 |
return rc; |
d01776393 mm: fix potential... |
1423 |
|
eb03aa008 mm/hugetlb: impro... |
1424 1425 1426 1427 1428 1429 1430 1431 |
for (pfn = start_pfn; pfn < end_pfn; pfn += 1 << minimum_order) { page = pfn_to_page(pfn); if (PageHuge(page) && !page_count(page)) { rc = dissolve_free_huge_page(page); if (rc) break; } } |
082d5b6b6 mm/hugetlb: check... |
1432 1433 |
return rc; |
c8721bbbd mm: memory-hotplu... |
1434 |
} |
099730d67 mm, hugetlb: use ... |
1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 |
/* * There are 3 ways this can get called: * 1. With vma+addr: we use the VMA's memory policy * 2. With !vma, but nid=NUMA_NO_NODE: We try to allocate a huge * page from any node, and let the buddy allocator itself figure * it out. * 3. With !vma, but nid!=NUMA_NO_NODE. We allocate a huge page * strictly from 'nid' */ static struct page *__hugetlb_alloc_buddy_huge_page(struct hstate *h, struct vm_area_struct *vma, unsigned long addr, int nid) { int order = huge_page_order(h); gfp_t gfp = htlb_alloc_mask(h)|__GFP_COMP|__GFP_REPEAT|__GFP_NOWARN; unsigned int cpuset_mems_cookie; /* * We need a VMA to get a memory policy. If we do not |
e0ec90ee7 mm, hugetlbfs: op... |
1453 1454 1455 1456 1457 1458 |
* have one, we use the 'nid' argument. * * The mempolicy stuff below has some non-inlined bits * and calls ->vm_ops. That makes it hard to optimize at * compile-time, even when NUMA is off and it does * nothing. This helps the compiler optimize it out. |
099730d67 mm, hugetlb: use ... |
1459 |
*/ |
e0ec90ee7 mm, hugetlbfs: op... |
1460 |
if (!IS_ENABLED(CONFIG_NUMA) || !vma) { |
099730d67 mm, hugetlb: use ... |
1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 |
/* * If a specific node is requested, make sure to * get memory from there, but only when a node * is explicitly specified. */ if (nid != NUMA_NO_NODE) gfp |= __GFP_THISNODE; /* * Make sure to call something that can handle * nid=NUMA_NO_NODE */ return alloc_pages_node(nid, gfp, order); } /* * OK, so we have a VMA. Fetch the mempolicy and try to |
e0ec90ee7 mm, hugetlbfs: op... |
1477 1478 |
* allocate a huge page with it. We will only reach this * when CONFIG_NUMA=y. |
099730d67 mm, hugetlb: use ... |
1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 |
*/ do { struct page *page; struct mempolicy *mpol; struct zonelist *zl; nodemask_t *nodemask; cpuset_mems_cookie = read_mems_allowed_begin(); zl = huge_zonelist(vma, addr, gfp, &mpol, &nodemask); mpol_cond_put(mpol); page = __alloc_pages_nodemask(gfp, order, zl, nodemask); if (page) return page; } while (read_mems_allowed_retry(cpuset_mems_cookie)); return NULL; } /* * There are two ways to allocate a huge page: * 1. When you have a VMA and an address (like a fault) * 2. When you have no VMA (like when setting /proc/.../nr_hugepages) * * 'vma' and 'addr' are only for (1). 'nid' is always NUMA_NO_NODE in * this case which signifies that the allocation should be done with * respect for the VMA's memory policy. * * For (2), we ignore 'vma' and 'addr' and use 'nid' exclusively. This * implies that memory policies will not be taken in to account. */ static struct page *__alloc_buddy_huge_page(struct hstate *h, struct vm_area_struct *vma, unsigned long addr, int nid) |
7893d1d50 hugetlb: Try to g... |
1511 1512 |
{ struct page *page; |
bf50bab2b hugetlb: add allo... |
1513 |
unsigned int r_nid; |
7893d1d50 hugetlb: Try to g... |
1514 |
|
bae7f4ae1 hugetlb: add hsta... |
1515 |
if (hstate_is_gigantic(h)) |
aa888a749 hugetlb: support ... |
1516 |
return NULL; |
d1c3fb1f8 hugetlb: introduc... |
1517 |
/* |
099730d67 mm, hugetlb: use ... |
1518 1519 1520 1521 1522 |
* Make sure that anyone specifying 'nid' is not also specifying a VMA. * This makes sure the caller is picking _one_ of the modes with which * we can call this function, not both. */ if (vma || (addr != -1)) { |
e0ec90ee7 mm, hugetlbfs: op... |
1523 1524 |
VM_WARN_ON_ONCE(addr == -1); VM_WARN_ON_ONCE(nid != NUMA_NO_NODE); |
099730d67 mm, hugetlb: use ... |
1525 1526 |
} /* |
d1c3fb1f8 hugetlb: introduc... |
1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 |
* Assume we will successfully allocate the surplus page to * prevent racing processes from causing the surplus to exceed * overcommit * * This however introduces a different race, where a process B * tries to grow the static hugepage pool while alloc_pages() is * called by process A. B will only examine the per-node * counters in determining if surplus huge pages can be * converted to normal huge pages in adjust_pool_surplus(). A * won't be able to increment the per-node counter, until the * lock is dropped by B, but B doesn't drop hugetlb_lock until * no more huge pages can be converted from surplus to normal * state (and doesn't try to convert again). Thus, we have a * case where a surplus huge page exists, the pool is grown, and * the surplus huge page still exists after, even though it * should just have been converted to a normal huge page. This * does not leak memory, though, as the hugepage will be freed * once it is out of use. It also does not allow the counters to * go out of whack in adjust_pool_surplus() as we don't modify * the node values until we've gotten the hugepage and only the * per-node value is checked there. */ spin_lock(&hugetlb_lock); |
a55164389 hugetlb: modular ... |
1550 |
if (h->surplus_huge_pages >= h->nr_overcommit_huge_pages) { |
d1c3fb1f8 hugetlb: introduc... |
1551 1552 1553 |
spin_unlock(&hugetlb_lock); return NULL; } else { |
a55164389 hugetlb: modular ... |
1554 1555 |
h->nr_huge_pages++; h->surplus_huge_pages++; |
d1c3fb1f8 hugetlb: introduc... |
1556 1557 |
} spin_unlock(&hugetlb_lock); |
099730d67 mm, hugetlb: use ... |
1558 |
page = __hugetlb_alloc_buddy_huge_page(h, vma, addr, nid); |
d1c3fb1f8 hugetlb: introduc... |
1559 1560 |
spin_lock(&hugetlb_lock); |
7893d1d50 hugetlb: Try to g... |
1561 |
if (page) { |
0edaecfab hugetlb: add a li... |
1562 |
INIT_LIST_HEAD(&page->lru); |
bf50bab2b hugetlb: add allo... |
1563 |
r_nid = page_to_nid(page); |
f1e61557f mm: pack compound... |
1564 |
set_compound_page_dtor(page, HUGETLB_PAGE_DTOR); |
9dd540e23 hugetlb/cgroup: a... |
1565 |
set_hugetlb_cgroup(page, NULL); |
d1c3fb1f8 hugetlb: introduc... |
1566 1567 1568 |
/* * We incremented the global counters already */ |
bf50bab2b hugetlb: add allo... |
1569 1570 |
h->nr_huge_pages_node[r_nid]++; h->surplus_huge_pages_node[r_nid]++; |
3b1163006 Subject: [PATCH] ... |
1571 |
__count_vm_event(HTLB_BUDDY_PGALLOC); |
d1c3fb1f8 hugetlb: introduc... |
1572 |
} else { |
a55164389 hugetlb: modular ... |
1573 1574 |
h->nr_huge_pages--; h->surplus_huge_pages--; |
3b1163006 Subject: [PATCH] ... |
1575 |
__count_vm_event(HTLB_BUDDY_PGALLOC_FAIL); |
7893d1d50 hugetlb: Try to g... |
1576 |
} |
d1c3fb1f8 hugetlb: introduc... |
1577 |
spin_unlock(&hugetlb_lock); |
7893d1d50 hugetlb: Try to g... |
1578 1579 1580 |
return page; } |
e4e574b76 hugetlb: Try to g... |
1581 |
/* |
099730d67 mm, hugetlb: use ... |
1582 1583 1584 1585 |
* Allocate a huge page from 'nid'. Note, 'nid' may be * NUMA_NO_NODE, which means that it may be allocated * anywhere. */ |
e0ec90ee7 mm, hugetlbfs: op... |
1586 |
static |
099730d67 mm, hugetlb: use ... |
1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 |
struct page *__alloc_buddy_huge_page_no_mpol(struct hstate *h, int nid) { unsigned long addr = -1; return __alloc_buddy_huge_page(h, NULL, addr, nid); } /* * Use the VMA's mpolicy to allocate a huge page from the buddy. */ |
e0ec90ee7 mm, hugetlbfs: op... |
1597 |
static |
099730d67 mm, hugetlb: use ... |
1598 1599 1600 1601 1602 1603 1604 |
struct page *__alloc_buddy_huge_page_with_mpol(struct hstate *h, struct vm_area_struct *vma, unsigned long addr) { return __alloc_buddy_huge_page(h, vma, addr, NUMA_NO_NODE); } /* |
bf50bab2b hugetlb: add allo... |
1605 1606 1607 1608 1609 1610 |
* This allocation function is useful in the context where vma is irrelevant. * E.g. soft-offlining uses this function because it only cares physical * address of error page. */ struct page *alloc_huge_page_node(struct hstate *h, int nid) { |
4ef918480 mm, hugetlb: prot... |
1611 |
struct page *page = NULL; |
bf50bab2b hugetlb: add allo... |
1612 1613 |
spin_lock(&hugetlb_lock); |
4ef918480 mm, hugetlb: prot... |
1614 1615 |
if (h->free_huge_pages - h->resv_huge_pages > 0) page = dequeue_huge_page_node(h, nid); |
bf50bab2b hugetlb: add allo... |
1616 |
spin_unlock(&hugetlb_lock); |
94ae8ba71 hugetlb/cgroup: a... |
1617 |
if (!page) |
099730d67 mm, hugetlb: use ... |
1618 |
page = __alloc_buddy_huge_page_no_mpol(h, nid); |
bf50bab2b hugetlb: add allo... |
1619 1620 1621 1622 1623 |
return page; } /* |
25985edce Fix common misspe... |
1624 |
* Increase the hugetlb pool such that it can accommodate a reservation |
e4e574b76 hugetlb: Try to g... |
1625 1626 |
* of size 'delta'. */ |
a55164389 hugetlb: modular ... |
1627 |
static int gather_surplus_pages(struct hstate *h, int delta) |
e4e574b76 hugetlb: Try to g... |
1628 1629 1630 1631 1632 |
{ struct list_head surplus_list; struct page *page, *tmp; int ret, i; int needed, allocated; |
28073b02b mm: hugetlb: defe... |
1633 |
bool alloc_ok = true; |
e4e574b76 hugetlb: Try to g... |
1634 |
|
a55164389 hugetlb: modular ... |
1635 |
needed = (h->resv_huge_pages + delta) - h->free_huge_pages; |
ac09b3a15 hugetlb: close a ... |
1636 |
if (needed <= 0) { |
a55164389 hugetlb: modular ... |
1637 |
h->resv_huge_pages += delta; |
e4e574b76 hugetlb: Try to g... |
1638 |
return 0; |
ac09b3a15 hugetlb: close a ... |
1639 |
} |
e4e574b76 hugetlb: Try to g... |
1640 1641 1642 1643 1644 1645 1646 1647 |
allocated = 0; INIT_LIST_HEAD(&surplus_list); ret = -ENOMEM; retry: spin_unlock(&hugetlb_lock); for (i = 0; i < needed; i++) { |
099730d67 mm, hugetlb: use ... |
1648 |
page = __alloc_buddy_huge_page_no_mpol(h, NUMA_NO_NODE); |
28073b02b mm: hugetlb: defe... |
1649 1650 1651 1652 |
if (!page) { alloc_ok = false; break; } |
e4e574b76 hugetlb: Try to g... |
1653 1654 |
list_add(&page->lru, &surplus_list); } |
28073b02b mm: hugetlb: defe... |
1655 |
allocated += i; |
e4e574b76 hugetlb: Try to g... |
1656 1657 1658 1659 1660 1661 |
/* * 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 ... |
1662 1663 |
needed = (h->resv_huge_pages + delta) - (h->free_huge_pages + allocated); |
28073b02b mm: hugetlb: defe... |
1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 |
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... |
1674 1675 |
/* * The surplus_list now contains _at_least_ the number of extra pages |
25985edce Fix common misspe... |
1676 |
* needed to accommodate the reservation. Add the appropriate number |
e4e574b76 hugetlb: Try to g... |
1677 |
* of pages to the hugetlb pool and free the extras back to the buddy |
ac09b3a15 hugetlb: close a ... |
1678 1679 1680 |
* 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... |
1681 1682 |
*/ needed += allocated; |
a55164389 hugetlb: modular ... |
1683 |
h->resv_huge_pages += delta; |
e4e574b76 hugetlb: Try to g... |
1684 |
ret = 0; |
a9869b837 hugetlb: move ref... |
1685 |
|
19fc3f0ac hugetlb: decrease... |
1686 |
/* Free the needed pages to the hugetlb pool */ |
e4e574b76 hugetlb: Try to g... |
1687 |
list_for_each_entry_safe(page, tmp, &surplus_list, lru) { |
19fc3f0ac hugetlb: decrease... |
1688 1689 |
if ((--needed) < 0) break; |
a9869b837 hugetlb: move ref... |
1690 1691 1692 1693 1694 |
/* * 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... |
1695 |
VM_BUG_ON_PAGE(page_count(page), page); |
a55164389 hugetlb: modular ... |
1696 |
enqueue_huge_page(h, page); |
19fc3f0ac hugetlb: decrease... |
1697 |
} |
28073b02b mm: hugetlb: defe... |
1698 |
free: |
b0365c8d0 mm: hugetlb: fix ... |
1699 |
spin_unlock(&hugetlb_lock); |
19fc3f0ac hugetlb: decrease... |
1700 1701 |
/* Free unnecessary surplus pages to the buddy allocator */ |
c0d934ba2 mm, hugetlb: remo... |
1702 1703 |
list_for_each_entry_safe(page, tmp, &surplus_list, lru) put_page(page); |
a9869b837 hugetlb: move ref... |
1704 |
spin_lock(&hugetlb_lock); |
e4e574b76 hugetlb: Try to g... |
1705 1706 1707 1708 1709 |
return ret; } /* |
1e26cec60 mm/hugetlb.c: fix... |
1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 |
* 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... |
1722 |
*/ |
a55164389 hugetlb: modular ... |
1723 1724 |
static void return_unused_surplus_pages(struct hstate *h, unsigned long unused_resv_pages) |
e4e574b76 hugetlb: Try to g... |
1725 |
{ |
e4e574b76 hugetlb: Try to g... |
1726 |
unsigned long nr_pages; |
aa888a749 hugetlb: support ... |
1727 |
/* Cannot return gigantic pages currently */ |
bae7f4ae1 hugetlb: add hsta... |
1728 |
if (hstate_is_gigantic(h)) |
1e26cec60 mm/hugetlb.c: fix... |
1729 |
goto out; |
aa888a749 hugetlb: support ... |
1730 |
|
1e26cec60 mm/hugetlb.c: fix... |
1731 1732 1733 1734 |
/* * Part (or even all) of the reservation could have been backed * by pre-allocated pages. Only free surplus pages. */ |
a55164389 hugetlb: modular ... |
1735 |
nr_pages = min(unused_resv_pages, h->surplus_huge_pages); |
e4e574b76 hugetlb: Try to g... |
1736 |
|
685f34570 hugetlb: use free... |
1737 1738 |
/* * We want to release as many surplus pages as possible, spread |
9b5e5d0fd hugetlb: use only... |
1739 1740 1741 1742 1743 |
* 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. |
1e26cec60 mm/hugetlb.c: fix... |
1744 1745 1746 1747 |
* * 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... |
1748 1749 |
*/ while (nr_pages--) { |
1e26cec60 mm/hugetlb.c: fix... |
1750 1751 |
h->resv_huge_pages--; unused_resv_pages--; |
8cebfcd07 hugetlb: use N_ME... |
1752 |
if (!free_pool_huge_page(h, &node_states[N_MEMORY], 1)) |
1e26cec60 mm/hugetlb.c: fix... |
1753 |
goto out; |
7848a4bf5 mm/hugetlb.c: add... |
1754 |
cond_resched_lock(&hugetlb_lock); |
e4e574b76 hugetlb: Try to g... |
1755 |
} |
1e26cec60 mm/hugetlb.c: fix... |
1756 1757 1758 1759 |
out: /* Fully uncommit the reservation */ h->resv_huge_pages -= unused_resv_pages; |
e4e574b76 hugetlb: Try to g... |
1760 |
} |
5e9113731 mm/hugetlb: add c... |
1761 |
|
c37f9fb11 hugetlb: allow hu... |
1762 |
/* |
feba16e25 mm/hugetlb: add r... |
1763 |
* vma_needs_reservation, vma_commit_reservation and vma_end_reservation |
5e9113731 mm/hugetlb: add c... |
1764 |
* are used by the huge page allocation routines to manage reservations. |
cf3ad20bf mm/hugetlb: compu... |
1765 1766 1767 1768 1769 1770 |
* * 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... |
1771 1772 1773 |
* 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... |
1774 1775 1776 1777 1778 1779 |
* * 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... |
1780 1781 1782 1783 1784 |
* * 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... |
1785 |
*/ |
5e9113731 mm/hugetlb: add c... |
1786 1787 1788 |
enum vma_resv_mode { VMA_NEEDS_RESV, VMA_COMMIT_RESV, |
feba16e25 mm/hugetlb: add r... |
1789 |
VMA_END_RESV, |
96b96a96d mm/hugetlb: fix h... |
1790 |
VMA_ADD_RESV, |
5e9113731 mm/hugetlb: add c... |
1791 |
}; |
cf3ad20bf mm/hugetlb: compu... |
1792 1793 |
static long __vma_reservation_common(struct hstate *h, struct vm_area_struct *vma, unsigned long addr, |
5e9113731 mm/hugetlb: add c... |
1794 |
enum vma_resv_mode mode) |
c37f9fb11 hugetlb: allow hu... |
1795 |
{ |
4e35f4838 mm, hugetlb: use ... |
1796 1797 |
struct resv_map *resv; pgoff_t idx; |
cf3ad20bf mm/hugetlb: compu... |
1798 |
long ret; |
c37f9fb11 hugetlb: allow hu... |
1799 |
|
4e35f4838 mm, hugetlb: use ... |
1800 1801 |
resv = vma_resv_map(vma); if (!resv) |
84afd99b8 hugetlb reservati... |
1802 |
return 1; |
c37f9fb11 hugetlb: allow hu... |
1803 |
|
4e35f4838 mm, hugetlb: use ... |
1804 |
idx = vma_hugecache_offset(h, vma, addr); |
5e9113731 mm/hugetlb: add c... |
1805 1806 |
switch (mode) { case VMA_NEEDS_RESV: |
cf3ad20bf mm/hugetlb: compu... |
1807 |
ret = region_chg(resv, idx, idx + 1); |
5e9113731 mm/hugetlb: add c... |
1808 1809 1810 1811 |
break; case VMA_COMMIT_RESV: ret = region_add(resv, idx, idx + 1); break; |
feba16e25 mm/hugetlb: add r... |
1812 |
case VMA_END_RESV: |
5e9113731 mm/hugetlb: add c... |
1813 1814 1815 |
region_abort(resv, idx, idx + 1); ret = 0; break; |
96b96a96d mm/hugetlb: fix h... |
1816 1817 1818 1819 1820 1821 1822 1823 |
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... |
1824 1825 1826 |
default: BUG(); } |
84afd99b8 hugetlb reservati... |
1827 |
|
4e35f4838 mm, hugetlb: use ... |
1828 |
if (vma->vm_flags & VM_MAYSHARE) |
cf3ad20bf mm/hugetlb: compu... |
1829 |
return ret; |
67961f9db mm/hugetlb: fix h... |
1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 |
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 ... |
1849 |
else |
cf3ad20bf mm/hugetlb: compu... |
1850 |
return ret < 0 ? ret : 0; |
c37f9fb11 hugetlb: allow hu... |
1851 |
} |
cf3ad20bf mm/hugetlb: compu... |
1852 1853 |
static long vma_needs_reservation(struct hstate *h, |
a55164389 hugetlb: modular ... |
1854 |
struct vm_area_struct *vma, unsigned long addr) |
c37f9fb11 hugetlb: allow hu... |
1855 |
{ |
5e9113731 mm/hugetlb: add c... |
1856 |
return __vma_reservation_common(h, vma, addr, VMA_NEEDS_RESV); |
cf3ad20bf mm/hugetlb: compu... |
1857 |
} |
84afd99b8 hugetlb reservati... |
1858 |
|
cf3ad20bf mm/hugetlb: compu... |
1859 1860 1861 |
static long vma_commit_reservation(struct hstate *h, struct vm_area_struct *vma, unsigned long addr) { |
5e9113731 mm/hugetlb: add c... |
1862 1863 |
return __vma_reservation_common(h, vma, addr, VMA_COMMIT_RESV); } |
feba16e25 mm/hugetlb: add r... |
1864 |
static void vma_end_reservation(struct hstate *h, |
5e9113731 mm/hugetlb: add c... |
1865 1866 |
struct vm_area_struct *vma, unsigned long addr) { |
feba16e25 mm/hugetlb: add r... |
1867 |
(void)__vma_reservation_common(h, vma, addr, VMA_END_RESV); |
c37f9fb11 hugetlb: allow hu... |
1868 |
} |
96b96a96d mm/hugetlb: fix h... |
1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 |
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... |
1918 |
struct page *alloc_huge_page(struct vm_area_struct *vma, |
04f2cbe35 hugetlb: guarante... |
1919 |
unsigned long addr, int avoid_reserve) |
1da177e4c Linux-2.6.12-rc2 |
1920 |
{ |
90481622d hugepages: fix us... |
1921 |
struct hugepage_subpool *spool = subpool_vma(vma); |
a55164389 hugetlb: modular ... |
1922 |
struct hstate *h = hstate_vma(vma); |
348ea204c hugetlb: split al... |
1923 |
struct page *page; |
d85f69b0b mm/hugetlb: alloc... |
1924 1925 |
long map_chg, map_commit; long gbl_chg; |
6d76dcf40 hugetlb/cgroup: a... |
1926 1927 |
int ret, idx; struct hugetlb_cgroup *h_cg; |
a1e78772d hugetlb: reserve ... |
1928 |
|
6d76dcf40 hugetlb/cgroup: a... |
1929 |
idx = hstate_index(h); |
a1e78772d hugetlb: reserve ... |
1930 |
/* |
d85f69b0b mm/hugetlb: alloc... |
1931 1932 1933 |
* 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 ... |
1934 |
*/ |
d85f69b0b mm/hugetlb: alloc... |
1935 1936 |
map_chg = gbl_chg = vma_needs_reservation(h, vma, addr); if (map_chg < 0) |
76dcee75c hugetlb: don't us... |
1937 |
return ERR_PTR(-ENOMEM); |
d85f69b0b mm/hugetlb: alloc... |
1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 |
/* * 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... |
1949 |
vma_end_reservation(h, vma, addr); |
76dcee75c hugetlb: don't us... |
1950 |
return ERR_PTR(-ENOSPC); |
5e9113731 mm/hugetlb: add c... |
1951 |
} |
1da177e4c Linux-2.6.12-rc2 |
1952 |
|
d85f69b0b mm/hugetlb: alloc... |
1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 |
/* * 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... |
1964 |
ret = hugetlb_cgroup_charge_cgroup(idx, pages_per_huge_page(h), &h_cg); |
8f34af6f9 mm, hugetlb: move... |
1965 1966 |
if (ret) goto out_subpool_put; |
1da177e4c Linux-2.6.12-rc2 |
1967 |
spin_lock(&hugetlb_lock); |
d85f69b0b mm/hugetlb: alloc... |
1968 1969 1970 1971 1972 1973 |
/* * 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... |
1974 |
if (!page) { |
94ae8ba71 hugetlb/cgroup: a... |
1975 |
spin_unlock(&hugetlb_lock); |
099730d67 mm, hugetlb: use ... |
1976 |
page = __alloc_buddy_huge_page_with_mpol(h, vma, addr); |
8f34af6f9 mm, hugetlb: move... |
1977 1978 |
if (!page) goto out_uncharge_cgroup; |
a88c76954 mm: hugetlb: fix ... |
1979 1980 1981 1982 |
if (!avoid_reserve && vma_has_reserves(vma, gbl_chg)) { SetPagePrivate(page); h->resv_huge_pages--; } |
79dbb2368 hugetlb: move all... |
1983 1984 |
spin_lock(&hugetlb_lock); list_move(&page->lru, &h->hugepage_activelist); |
81a6fcae3 mm, hugetlb: clea... |
1985 |
/* Fall through */ |
68842c9b9 hugetlbfs: fix qu... |
1986 |
} |
81a6fcae3 mm, hugetlb: clea... |
1987 1988 |
hugetlb_cgroup_commit_charge(idx, pages_per_huge_page(h), h_cg, page); spin_unlock(&hugetlb_lock); |
348ea204c hugetlb: split al... |
1989 |
|
90481622d hugepages: fix us... |
1990 |
set_page_private(page, (unsigned long)spool); |
90d8b7e61 hugetlb: enforce ... |
1991 |
|
d85f69b0b mm/hugetlb: alloc... |
1992 1993 |
map_commit = vma_commit_reservation(h, vma, addr); if (unlikely(map_chg > map_commit)) { |
33039678c mm/hugetlb: handl... |
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 |
/* * 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 ... |
2008 |
return page; |
8f34af6f9 mm, hugetlb: move... |
2009 2010 2011 2012 |
out_uncharge_cgroup: hugetlb_cgroup_uncharge_cgroup(idx, pages_per_huge_page(h), h_cg); out_subpool_put: |
d85f69b0b mm/hugetlb: alloc... |
2013 |
if (map_chg || avoid_reserve) |
8f34af6f9 mm, hugetlb: move... |
2014 |
hugepage_subpool_put_pages(spool, 1); |
feba16e25 mm/hugetlb: add r... |
2015 |
vma_end_reservation(h, vma, addr); |
8f34af6f9 mm, hugetlb: move... |
2016 |
return ERR_PTR(-ENOSPC); |
b45b5bd65 [PATCH] hugepage:... |
2017 |
} |
74060e4d7 mm: mbind: add hu... |
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 |
/* * alloc_huge_page()'s wrapper which simply returns the page if allocation * succeeds, otherwise NULL. This function is called from new_vma_page(), * where no ERR_VALUE is expected to be returned. */ struct page *alloc_huge_page_noerr(struct vm_area_struct *vma, unsigned long addr, int avoid_reserve) { struct page *page = alloc_huge_page(vma, addr, avoid_reserve); if (IS_ERR(page)) page = NULL; return page; } |
91f47662d mm: hugetlb: remo... |
2031 |
int __weak alloc_bootmem_huge_page(struct hstate *h) |
aa888a749 hugetlb: support ... |
2032 2033 |
{ struct huge_bootmem_page *m; |
b22610268 mm, hugetlb: fix ... |
2034 |
int nr_nodes, node; |
aa888a749 hugetlb: support ... |
2035 |
|
b22610268 mm, hugetlb: fix ... |
2036 |
for_each_node_mask_to_alloc(h, nr_nodes, node, &node_states[N_MEMORY]) { |
aa888a749 hugetlb: support ... |
2037 |
void *addr; |
8b89a1169 mm/hugetlb.c: use... |
2038 2039 2040 |
addr = memblock_virt_alloc_try_nid_nopanic( huge_page_size(h), huge_page_size(h), 0, BOOTMEM_ALLOC_ACCESSIBLE, node); |
aa888a749 hugetlb: support ... |
2041 2042 2043 2044 2045 2046 2047 |
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... |
2048 |
goto found; |
aa888a749 hugetlb: support ... |
2049 |
} |
aa888a749 hugetlb: support ... |
2050 2051 2052 2053 |
} return 0; found: |
df994ead5 hugetlb: alloc_bo... |
2054 |
BUG_ON(!IS_ALIGNED(virt_to_phys(m), huge_page_size(h))); |
aa888a749 hugetlb: support ... |
2055 2056 2057 2058 2059 |
/* Put them into a private list first because mem_map is not up yet */ list_add(&m->list, &huge_boot_pages); m->hstate = h; return 1; } |
d00181b96 mm: use 'unsigned... |
2060 2061 |
static void __init prep_compound_huge_page(struct page *page, unsigned int order) |
18229df5b hugetlb: pull gig... |
2062 2063 2064 2065 2066 2067 |
{ if (unlikely(order > (MAX_ORDER - 1))) prep_compound_gigantic_page(page, order); else prep_compound_page(page, order); } |
aa888a749 hugetlb: support ... |
2068 2069 2070 2071 2072 2073 |
/* Put bootmem huge pages into the standard lists after mem_map is up */ static void __init gather_bootmem_prealloc(void) { struct huge_bootmem_page *m; list_for_each_entry(m, &huge_boot_pages, list) { |
aa888a749 hugetlb: support ... |
2074 |
struct hstate *h = m->hstate; |
ee8f248d2 hugetlb: add phys... |
2075 2076 2077 2078 |
struct page *page; #ifdef CONFIG_HIGHMEM page = pfn_to_page(m->phys >> PAGE_SHIFT); |
8b89a1169 mm/hugetlb.c: use... |
2079 2080 |
memblock_free_late(__pa(m), sizeof(struct huge_bootmem_page)); |
ee8f248d2 hugetlb: add phys... |
2081 2082 2083 |
#else page = virt_to_page(m); #endif |
aa888a749 hugetlb: support ... |
2084 |
WARN_ON(page_count(page) != 1); |
18229df5b hugetlb: pull gig... |
2085 |
prep_compound_huge_page(page, h->order); |
ef5a22be2 mm: hugetlb: init... |
2086 |
WARN_ON(PageReserved(page)); |
aa888a749 hugetlb: support ... |
2087 |
prep_new_huge_page(h, page, page_to_nid(page)); |
b0320c7b7 mm: fix negative ... |
2088 2089 2090 2091 2092 2093 |
/* * 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... |
2094 |
if (hstate_is_gigantic(h)) |
3dcc0571c mm: correctly upd... |
2095 |
adjust_managed_page_count(page, 1 << h->order); |
aa888a749 hugetlb: support ... |
2096 2097 |
} } |
8faa8b077 hugetlb: support ... |
2098 |
static void __init hugetlb_hstate_alloc_pages(struct hstate *h) |
1da177e4c Linux-2.6.12-rc2 |
2099 2100 |
{ unsigned long i; |
a55164389 hugetlb: modular ... |
2101 |
|
e5ff21594 hugetlb: multiple... |
2102 |
for (i = 0; i < h->max_huge_pages; ++i) { |
bae7f4ae1 hugetlb: add hsta... |
2103 |
if (hstate_is_gigantic(h)) { |
aa888a749 hugetlb: support ... |
2104 2105 |
if (!alloc_bootmem_huge_page(h)) break; |
9b5e5d0fd hugetlb: use only... |
2106 |
} else if (!alloc_fresh_huge_page(h, |
8cebfcd07 hugetlb: use N_ME... |
2107 |
&node_states[N_MEMORY])) |
1da177e4c Linux-2.6.12-rc2 |
2108 |
break; |
1da177e4c Linux-2.6.12-rc2 |
2109 |
} |
8faa8b077 hugetlb: support ... |
2110 |
h->max_huge_pages = i; |
e5ff21594 hugetlb: multiple... |
2111 2112 2113 2114 2115 2116 2117 |
} static void __init hugetlb_init_hstates(void) { struct hstate *h; for_each_hstate(h) { |
641844f56 mm/hugetlb: intro... |
2118 2119 |
if (minimum_order > huge_page_order(h)) minimum_order = huge_page_order(h); |
8faa8b077 hugetlb: support ... |
2120 |
/* oversize hugepages were init'ed in early boot */ |
bae7f4ae1 hugetlb: add hsta... |
2121 |
if (!hstate_is_gigantic(h)) |
8faa8b077 hugetlb: support ... |
2122 |
hugetlb_hstate_alloc_pages(h); |
e5ff21594 hugetlb: multiple... |
2123 |
} |
641844f56 mm/hugetlb: intro... |
2124 |
VM_BUG_ON(minimum_order == UINT_MAX); |
e5ff21594 hugetlb: multiple... |
2125 |
} |
4abd32dba hugetlb: printk c... |
2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 |
static char * __init memfmt(char *buf, unsigned long n) { if (n >= (1UL << 30)) sprintf(buf, "%lu GB", n >> 30); else if (n >= (1UL << 20)) sprintf(buf, "%lu MB", n >> 20); else sprintf(buf, "%lu KB", n >> 10); return buf; } |
e5ff21594 hugetlb: multiple... |
2136 2137 2138 2139 2140 |
static void __init report_hugepages(void) { struct hstate *h; for_each_hstate(h) { |
4abd32dba hugetlb: printk c... |
2141 |
char buf[32]; |
ffb22af5b mm/hugetlb.c: con... |
2142 2143 |
pr_info("HugeTLB registered %s page size, pre-allocated %ld pages ", |
4abd32dba hugetlb: printk c... |
2144 2145 |
memfmt(buf, huge_page_size(h)), h->free_huge_pages); |
e5ff21594 hugetlb: multiple... |
2146 2147 |
} } |
1da177e4c Linux-2.6.12-rc2 |
2148 |
#ifdef CONFIG_HIGHMEM |
6ae11b278 hugetlb: add node... |
2149 2150 |
static void try_to_free_low(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c Linux-2.6.12-rc2 |
2151 |
{ |
4415cc8df [PATCH] Hugepages... |
2152 |
int i; |
bae7f4ae1 hugetlb: add hsta... |
2153 |
if (hstate_is_gigantic(h)) |
aa888a749 hugetlb: support ... |
2154 |
return; |
6ae11b278 hugetlb: add node... |
2155 |
for_each_node_mask(i, *nodes_allowed) { |
1da177e4c Linux-2.6.12-rc2 |
2156 |
struct page *page, *next; |
a55164389 hugetlb: modular ... |
2157 2158 2159 |
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... |
2160 |
return; |
1da177e4c Linux-2.6.12-rc2 |
2161 2162 2163 |
if (PageHighMem(page)) continue; list_del(&page->lru); |
e5ff21594 hugetlb: multiple... |
2164 |
update_and_free_page(h, page); |
a55164389 hugetlb: modular ... |
2165 2166 |
h->free_huge_pages--; h->free_huge_pages_node[page_to_nid(page)]--; |
1da177e4c Linux-2.6.12-rc2 |
2167 2168 2169 2170 |
} } } #else |
6ae11b278 hugetlb: add node... |
2171 2172 |
static inline void try_to_free_low(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c Linux-2.6.12-rc2 |
2173 2174 2175 |
{ } #endif |
20a0307c0 mm: introduce Pag... |
2176 2177 2178 2179 2180 |
/* * 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... |
2181 2182 |
static int adjust_pool_surplus(struct hstate *h, nodemask_t *nodes_allowed, int delta) |
20a0307c0 mm: introduce Pag... |
2183 |
{ |
b22610268 mm, hugetlb: fix ... |
2184 |
int nr_nodes, node; |
20a0307c0 mm: introduce Pag... |
2185 2186 |
VM_BUG_ON(delta != -1 && delta != 1); |
20a0307c0 mm: introduce Pag... |
2187 |
|
b22610268 mm, hugetlb: fix ... |
2188 2189 2190 2191 |
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 ... |
2192 |
} |
b22610268 mm, hugetlb: fix ... |
2193 2194 2195 2196 2197 |
} 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 ... |
2198 |
} |
b22610268 mm, hugetlb: fix ... |
2199 2200 |
} return 0; |
20a0307c0 mm: introduce Pag... |
2201 |
|
b22610268 mm, hugetlb: fix ... |
2202 2203 2204 2205 |
found: h->surplus_huge_pages += delta; h->surplus_huge_pages_node[node] += delta; return 1; |
20a0307c0 mm: introduce Pag... |
2206 |
} |
a55164389 hugetlb: modular ... |
2207 |
#define persistent_huge_pages(h) (h->nr_huge_pages - h->surplus_huge_pages) |
6ae11b278 hugetlb: add node... |
2208 2209 |
static unsigned long set_max_huge_pages(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c Linux-2.6.12-rc2 |
2210 |
{ |
7893d1d50 hugetlb: Try to g... |
2211 |
unsigned long min_count, ret; |
1da177e4c Linux-2.6.12-rc2 |
2212 |
|
944d9fec8 hugetlb: add supp... |
2213 |
if (hstate_is_gigantic(h) && !gigantic_page_supported()) |
aa888a749 hugetlb: support ... |
2214 |
return h->max_huge_pages; |
7893d1d50 hugetlb: Try to g... |
2215 2216 2217 2218 |
/* * 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... |
2219 |
* |
d15c7c093 hugetlb: trivial ... |
2220 |
* We might race with __alloc_buddy_huge_page() here and be unable |
d1c3fb1f8 hugetlb: introduc... |
2221 2222 2223 2224 |
* 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... |
2225 |
*/ |
1da177e4c Linux-2.6.12-rc2 |
2226 |
spin_lock(&hugetlb_lock); |
a55164389 hugetlb: modular ... |
2227 |
while (h->surplus_huge_pages && count > persistent_huge_pages(h)) { |
6ae11b278 hugetlb: add node... |
2228 |
if (!adjust_pool_surplus(h, nodes_allowed, -1)) |
7893d1d50 hugetlb: Try to g... |
2229 2230 |
break; } |
a55164389 hugetlb: modular ... |
2231 |
while (count > persistent_huge_pages(h)) { |
7893d1d50 hugetlb: Try to g... |
2232 2233 2234 2235 2236 2237 |
/* * 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... |
2238 2239 2240 |
/* yield cpu to avoid soft lockup */ cond_resched(); |
944d9fec8 hugetlb: add supp... |
2241 2242 2243 2244 |
if (hstate_is_gigantic(h)) ret = alloc_fresh_gigantic_page(h, nodes_allowed); else ret = alloc_fresh_huge_page(h, nodes_allowed); |
7893d1d50 hugetlb: Try to g... |
2245 2246 2247 |
spin_lock(&hugetlb_lock); if (!ret) goto out; |
536240f2b hugetlb: abort a ... |
2248 2249 2250 |
/* Bail for signals. Probably ctrl-c from user */ if (signal_pending(current)) goto out; |
7893d1d50 hugetlb: Try to g... |
2251 |
} |
7893d1d50 hugetlb: Try to g... |
2252 2253 2254 2255 2256 2257 2258 |
/* * 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... |
2259 2260 2261 2262 |
* * 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 |
d15c7c093 hugetlb: trivial ... |
2263 |
* __alloc_buddy_huge_page() is checking the global counter, |
d1c3fb1f8 hugetlb: introduc... |
2264 2265 2266 |
* 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... |
2267 |
*/ |
a55164389 hugetlb: modular ... |
2268 |
min_count = h->resv_huge_pages + h->nr_huge_pages - h->free_huge_pages; |
6b0c880df hugetlb: fix pool... |
2269 |
min_count = max(count, min_count); |
6ae11b278 hugetlb: add node... |
2270 |
try_to_free_low(h, min_count, nodes_allowed); |
a55164389 hugetlb: modular ... |
2271 |
while (min_count < persistent_huge_pages(h)) { |
6ae11b278 hugetlb: add node... |
2272 |
if (!free_pool_huge_page(h, nodes_allowed, 0)) |
1da177e4c Linux-2.6.12-rc2 |
2273 |
break; |
55f67141a mm: hugetlb: fix ... |
2274 |
cond_resched_lock(&hugetlb_lock); |
1da177e4c Linux-2.6.12-rc2 |
2275 |
} |
a55164389 hugetlb: modular ... |
2276 |
while (count < persistent_huge_pages(h)) { |
6ae11b278 hugetlb: add node... |
2277 |
if (!adjust_pool_surplus(h, nodes_allowed, 1)) |
7893d1d50 hugetlb: Try to g... |
2278 2279 2280 |
break; } out: |
a55164389 hugetlb: modular ... |
2281 |
ret = persistent_huge_pages(h); |
1da177e4c Linux-2.6.12-rc2 |
2282 |
spin_unlock(&hugetlb_lock); |
7893d1d50 hugetlb: Try to g... |
2283 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
2284 |
} |
a34378701 hugetlb: new sysf... |
2285 2286 2287 2288 2289 2290 2291 2292 2293 |
#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 ... |
2294 2295 2296 |
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... |
2297 2298 |
{ int i; |
9a3052306 hugetlb: add per ... |
2299 |
|
a34378701 hugetlb: new sysf... |
2300 |
for (i = 0; i < HUGE_MAX_HSTATE; i++) |
9a3052306 hugetlb: add per ... |
2301 2302 2303 |
if (hstate_kobjs[i] == kobj) { if (nidp) *nidp = NUMA_NO_NODE; |
a34378701 hugetlb: new sysf... |
2304 |
return &hstates[i]; |
9a3052306 hugetlb: add per ... |
2305 2306 2307 |
} return kobj_to_node_hstate(kobj, nidp); |
a34378701 hugetlb: new sysf... |
2308 |
} |
06808b082 hugetlb: derive h... |
2309 |
static ssize_t nr_hugepages_show_common(struct kobject *kobj, |
a34378701 hugetlb: new sysf... |
2310 2311 |
struct kobj_attribute *attr, char *buf) { |
9a3052306 hugetlb: add per ... |
2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 |
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... |
2324 |
} |
adbe8726d hugetlb: do not a... |
2325 |
|
238d3c13f mm, hugetlb: gene... |
2326 2327 2328 |
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... |
2329 2330 |
{ int err; |
bad44b5be mm: add gfp flags... |
2331 |
NODEMASK_ALLOC(nodemask_t, nodes_allowed, GFP_KERNEL | __GFP_NORETRY); |
a34378701 hugetlb: new sysf... |
2332 |
|
944d9fec8 hugetlb: add supp... |
2333 |
if (hstate_is_gigantic(h) && !gigantic_page_supported()) { |
adbe8726d hugetlb: do not a... |
2334 2335 2336 |
err = -EINVAL; goto out; } |
9a3052306 hugetlb: add per ... |
2337 2338 2339 2340 2341 2342 2343 |
if (nid == NUMA_NO_NODE) { /* * global hstate attribute */ if (!(obey_mempolicy && init_nodemask_of_mempolicy(nodes_allowed))) { NODEMASK_FREE(nodes_allowed); |
8cebfcd07 hugetlb: use N_ME... |
2344 |
nodes_allowed = &node_states[N_MEMORY]; |
9a3052306 hugetlb: add per ... |
2345 2346 2347 2348 2349 2350 2351 2352 2353 |
} } else if (nodes_allowed) { /* * per node hstate attribute: adjust count to global, * but restrict alloc/free to the specified node. */ count += h->nr_huge_pages - h->nr_huge_pages_node[nid]; init_nodemask_of_node(nodes_allowed, nid); } else |
8cebfcd07 hugetlb: use N_ME... |
2354 |
nodes_allowed = &node_states[N_MEMORY]; |
9a3052306 hugetlb: add per ... |
2355 |
|
06808b082 hugetlb: derive h... |
2356 |
h->max_huge_pages = set_max_huge_pages(h, count, nodes_allowed); |
a34378701 hugetlb: new sysf... |
2357 |
|
8cebfcd07 hugetlb: use N_ME... |
2358 |
if (nodes_allowed != &node_states[N_MEMORY]) |
06808b082 hugetlb: derive h... |
2359 2360 2361 |
NODEMASK_FREE(nodes_allowed); return len; |
adbe8726d hugetlb: do not a... |
2362 2363 2364 |
out: NODEMASK_FREE(nodes_allowed); return err; |
06808b082 hugetlb: derive h... |
2365 |
} |
238d3c13f mm, hugetlb: gene... |
2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 |
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... |
2382 2383 2384 2385 2386 2387 2388 2389 2390 |
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... |
2391 |
return nr_hugepages_store_common(false, kobj, buf, len); |
a34378701 hugetlb: new sysf... |
2392 2393 |
} HSTATE_ATTR(nr_hugepages); |
06808b082 hugetlb: derive h... |
2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 |
#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... |
2409 |
return nr_hugepages_store_common(true, kobj, buf, len); |
06808b082 hugetlb: derive h... |
2410 2411 2412 |
} HSTATE_ATTR(nr_hugepages_mempolicy); #endif |
a34378701 hugetlb: new sysf... |
2413 2414 2415 |
static ssize_t nr_overcommit_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
9a3052306 hugetlb: add per ... |
2416 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701 hugetlb: new sysf... |
2417 2418 2419 |
return sprintf(buf, "%lu ", h->nr_overcommit_huge_pages); } |
adbe8726d hugetlb: do not a... |
2420 |
|
a34378701 hugetlb: new sysf... |
2421 2422 2423 2424 2425 |
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 ... |
2426 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701 hugetlb: new sysf... |
2427 |
|
bae7f4ae1 hugetlb: add hsta... |
2428 |
if (hstate_is_gigantic(h)) |
adbe8726d hugetlb: do not a... |
2429 |
return -EINVAL; |
3dbb95f78 mm: replace stric... |
2430 |
err = kstrtoul(buf, 10, &input); |
a34378701 hugetlb: new sysf... |
2431 |
if (err) |
73ae31e59 hugetlb: fix hand... |
2432 |
return err; |
a34378701 hugetlb: new sysf... |
2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 |
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 ... |
2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 |
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... |
2457 2458 2459 2460 2461 2462 |
} HSTATE_ATTR_RO(free_hugepages); static ssize_t resv_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
9a3052306 hugetlb: add per ... |
2463 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701 hugetlb: new sysf... |
2464 2465 2466 2467 2468 2469 2470 2471 |
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 ... |
2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 |
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... |
2484 2485 2486 2487 2488 2489 2490 2491 2492 |
} 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... |
2493 2494 2495 |
#ifdef CONFIG_NUMA &nr_hugepages_mempolicy_attr.attr, #endif |
a34378701 hugetlb: new sysf... |
2496 2497 2498 2499 2500 2501 |
NULL, }; static struct attribute_group hstate_attr_group = { .attrs = hstate_attrs, }; |
094e9539b hugetlb: fix sect... |
2502 2503 2504 |
static int hugetlb_sysfs_add_hstate(struct hstate *h, struct kobject *parent, struct kobject **hstate_kobjs, struct attribute_group *hstate_attr_group) |
a34378701 hugetlb: new sysf... |
2505 2506 |
{ int retval; |
972dc4de1 hugetlb: add an i... |
2507 |
int hi = hstate_index(h); |
a34378701 hugetlb: new sysf... |
2508 |
|
9a3052306 hugetlb: add per ... |
2509 2510 |
hstate_kobjs[hi] = kobject_create_and_add(h->name, parent); if (!hstate_kobjs[hi]) |
a34378701 hugetlb: new sysf... |
2511 |
return -ENOMEM; |
9a3052306 hugetlb: add per ... |
2512 |
retval = sysfs_create_group(hstate_kobjs[hi], hstate_attr_group); |
a34378701 hugetlb: new sysf... |
2513 |
if (retval) |
9a3052306 hugetlb: add per ... |
2514 |
kobject_put(hstate_kobjs[hi]); |
a34378701 hugetlb: new sysf... |
2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 |
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 ... |
2529 2530 |
err = hugetlb_sysfs_add_hstate(h, hugepages_kobj, hstate_kobjs, &hstate_attr_group); |
a34378701 hugetlb: new sysf... |
2531 |
if (err) |
ffb22af5b mm/hugetlb.c: con... |
2532 |
pr_err("Hugetlb: Unable to add hstate %s", h->name); |
a34378701 hugetlb: new sysf... |
2533 2534 |
} } |
9a3052306 hugetlb: add per ... |
2535 2536 2537 2538 |
#ifdef CONFIG_NUMA /* * node_hstate/s - associate per node hstate attributes, via their kobjects, |
10fbcf4c6 convert 'memory' ... |
2539 2540 2541 |
* 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 ... |
2542 2543 2544 2545 2546 2547 |
* 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 ... |
2548 |
static struct node_hstate node_hstates[MAX_NUMNODES]; |
9a3052306 hugetlb: add per ... |
2549 2550 |
/* |
10fbcf4c6 convert 'memory' ... |
2551 |
* A subset of global hstate attributes for node devices |
9a3052306 hugetlb: add per ... |
2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 |
*/ static struct attribute *per_node_hstate_attrs[] = { &nr_hugepages_attr.attr, &free_hugepages_attr.attr, &surplus_hugepages_attr.attr, NULL, }; static struct attribute_group per_node_hstate_attr_group = { .attrs = per_node_hstate_attrs, }; /* |
10fbcf4c6 convert 'memory' ... |
2565 |
* kobj_to_node_hstate - lookup global hstate for node device hstate attr kobj. |
9a3052306 hugetlb: add per ... |
2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 |
* 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' ... |
2588 |
* Unregister hstate attributes from a single node device. |
9a3052306 hugetlb: add per ... |
2589 2590 |
* No-op if no hstate attributes attached. */ |
3cd8b44fa hugetlb: fix spar... |
2591 |
static void hugetlb_unregister_node(struct node *node) |
9a3052306 hugetlb: add per ... |
2592 2593 |
{ struct hstate *h; |
10fbcf4c6 convert 'memory' ... |
2594 |
struct node_hstate *nhs = &node_hstates[node->dev.id]; |
9a3052306 hugetlb: add per ... |
2595 2596 |
if (!nhs->hugepages_kobj) |
9b5e5d0fd hugetlb: use only... |
2597 |
return; /* no hstate attributes */ |
9a3052306 hugetlb: add per ... |
2598 |
|
972dc4de1 hugetlb: add an i... |
2599 2600 2601 2602 2603 |
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 ... |
2604 |
} |
972dc4de1 hugetlb: add an i... |
2605 |
} |
9a3052306 hugetlb: add per ... |
2606 2607 2608 2609 |
kobject_put(nhs->hugepages_kobj); nhs->hugepages_kobj = NULL; } |
9a3052306 hugetlb: add per ... |
2610 2611 |
/* |
10fbcf4c6 convert 'memory' ... |
2612 |
* Register hstate attributes for a single node device. |
9a3052306 hugetlb: add per ... |
2613 2614 |
* No-op if attributes already registered. */ |
3cd8b44fa hugetlb: fix spar... |
2615 |
static void hugetlb_register_node(struct node *node) |
9a3052306 hugetlb: add per ... |
2616 2617 |
{ struct hstate *h; |
10fbcf4c6 convert 'memory' ... |
2618 |
struct node_hstate *nhs = &node_hstates[node->dev.id]; |
9a3052306 hugetlb: add per ... |
2619 2620 2621 2622 2623 2624 |
int err; if (nhs->hugepages_kobj) return; /* already allocated */ nhs->hugepages_kobj = kobject_create_and_add("hugepages", |
10fbcf4c6 convert 'memory' ... |
2625 |
&node->dev.kobj); |
9a3052306 hugetlb: add per ... |
2626 2627 2628 2629 2630 2631 2632 2633 |
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... |
2634 2635 2636 |
pr_err("Hugetlb: Unable to add hstate %s for node %d ", h->name, node->dev.id); |
9a3052306 hugetlb: add per ... |
2637 2638 2639 2640 2641 2642 2643 |
hugetlb_unregister_node(node); break; } } } /* |
9b5e5d0fd hugetlb: use only... |
2644 |
* hugetlb init time: register hstate attributes for all registered node |
10fbcf4c6 convert 'memory' ... |
2645 2646 |
* devices of nodes that have memory. All on-line nodes should have * registered their associated device by this time. |
9a3052306 hugetlb: add per ... |
2647 |
*/ |
7d9ca0004 hugetlb: hugetlb_... |
2648 |
static void __init hugetlb_register_all_nodes(void) |
9a3052306 hugetlb: add per ... |
2649 2650 |
{ int nid; |
8cebfcd07 hugetlb: use N_ME... |
2651 |
for_each_node_state(nid, N_MEMORY) { |
8732794b1 numa: convert sta... |
2652 |
struct node *node = node_devices[nid]; |
10fbcf4c6 convert 'memory' ... |
2653 |
if (node->dev.id == nid) |
9a3052306 hugetlb: add per ... |
2654 2655 2656 2657 |
hugetlb_register_node(node); } /* |
10fbcf4c6 convert 'memory' ... |
2658 |
* Let the node device driver know we're here so it can |
9a3052306 hugetlb: add per ... |
2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 |
* [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 ... |
2673 2674 2675 |
static void hugetlb_register_all_nodes(void) { } #endif |
a34378701 hugetlb: new sysf... |
2676 2677 |
static int __init hugetlb_init(void) { |
8382d914e mm, hugetlb: impr... |
2678 |
int i; |
457c1b27e hugetlb: ensure h... |
2679 |
if (!hugepages_supported()) |
0ef89d25d mm/hugetlb: don't... |
2680 |
return 0; |
a34378701 hugetlb: new sysf... |
2681 |
|
e11bfbfcb hugetlb: override... |
2682 2683 2684 2685 |
if (!size_to_hstate(default_hstate_size)) { default_hstate_size = HPAGE_SIZE; if (!size_to_hstate(default_hstate_size)) hugetlb_add_hstate(HUGETLB_PAGE_ORDER); |
a34378701 hugetlb: new sysf... |
2686 |
} |
972dc4de1 hugetlb: add an i... |
2687 |
default_hstate_idx = hstate_index(size_to_hstate(default_hstate_size)); |
f8b74815a mm/hugetlb.c: fix... |
2688 2689 2690 2691 |
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... |
2692 2693 |
hugetlb_init_hstates(); |
aa888a749 hugetlb: support ... |
2694 |
gather_bootmem_prealloc(); |
a34378701 hugetlb: new sysf... |
2695 2696 2697 |
report_hugepages(); hugetlb_sysfs_init(); |
9a3052306 hugetlb: add per ... |
2698 |
hugetlb_register_all_nodes(); |
7179e7bf4 mm/hugetlb: creat... |
2699 |
hugetlb_cgroup_file_init(); |
9a3052306 hugetlb: add per ... |
2700 |
|
8382d914e mm, hugetlb: impr... |
2701 2702 2703 2704 2705 |
#ifdef CONFIG_SMP num_fault_mutexes = roundup_pow_of_two(8 * num_possible_cpus()); #else num_fault_mutexes = 1; #endif |
c672c7f29 mm/hugetlb: expos... |
2706 |
hugetlb_fault_mutex_table = |
8382d914e mm, hugetlb: impr... |
2707 |
kmalloc(sizeof(struct mutex) * num_fault_mutexes, GFP_KERNEL); |
c672c7f29 mm/hugetlb: expos... |
2708 |
BUG_ON(!hugetlb_fault_mutex_table); |
8382d914e mm, hugetlb: impr... |
2709 2710 |
for (i = 0; i < num_fault_mutexes; i++) |
c672c7f29 mm/hugetlb: expos... |
2711 |
mutex_init(&hugetlb_fault_mutex_table[i]); |
a34378701 hugetlb: new sysf... |
2712 2713 |
return 0; } |
3e89e1c5e hugetlb: make mm ... |
2714 |
subsys_initcall(hugetlb_init); |
a34378701 hugetlb: new sysf... |
2715 2716 |
/* Should be called on processing a hugepagesz=... option */ |
9fee021d1 mm/hugetlb: intro... |
2717 2718 2719 2720 |
void __init hugetlb_bad_size(void) { parsed_valid_hugepagesz = false; } |
d00181b96 mm: use 'unsigned... |
2721 |
void __init hugetlb_add_hstate(unsigned int order) |
a34378701 hugetlb: new sysf... |
2722 2723 |
{ struct hstate *h; |
8faa8b077 hugetlb: support ... |
2724 |
unsigned long i; |
a34378701 hugetlb: new sysf... |
2725 |
if (size_to_hstate(PAGE_SIZE << order)) { |
598d80914 mm: convert pr_wa... |
2726 2727 |
pr_warn("hugepagesz= specified twice, ignoring "); |
a34378701 hugetlb: new sysf... |
2728 2729 |
return; } |
47d38344a hugetlb: rename m... |
2730 |
BUG_ON(hugetlb_max_hstate >= HUGE_MAX_HSTATE); |
a34378701 hugetlb: new sysf... |
2731 |
BUG_ON(order == 0); |
47d38344a hugetlb: rename m... |
2732 |
h = &hstates[hugetlb_max_hstate++]; |
a34378701 hugetlb: new sysf... |
2733 2734 |
h->order = order; h->mask = ~((1ULL << (order + PAGE_SHIFT)) - 1); |
8faa8b077 hugetlb: support ... |
2735 2736 2737 2738 |
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... |
2739 |
INIT_LIST_HEAD(&h->hugepage_activelist); |
54f18d352 mm/hugetlb.c: use... |
2740 2741 |
h->next_nid_to_alloc = first_memory_node; h->next_nid_to_free = first_memory_node; |
a34378701 hugetlb: new sysf... |
2742 2743 |
snprintf(h->name, HSTATE_NAME_LEN, "hugepages-%lukB", huge_page_size(h)/1024); |
8faa8b077 hugetlb: support ... |
2744 |
|
a34378701 hugetlb: new sysf... |
2745 2746 |
parsed_hstate = h; } |
e11bfbfcb hugetlb: override... |
2747 |
static int __init hugetlb_nrpages_setup(char *s) |
a34378701 hugetlb: new sysf... |
2748 2749 |
{ unsigned long *mhp; |
8faa8b077 hugetlb: support ... |
2750 |
static unsigned long *last_mhp; |
a34378701 hugetlb: new sysf... |
2751 |
|
9fee021d1 mm/hugetlb: intro... |
2752 2753 2754 2755 2756 2757 2758 |
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... |
2759 |
/* |
47d38344a hugetlb: rename m... |
2760 |
* !hugetlb_max_hstate means we haven't parsed a hugepagesz= parameter yet, |
a34378701 hugetlb: new sysf... |
2761 2762 |
* so this hugepages= parameter goes to the "default hstate". */ |
9fee021d1 mm/hugetlb: intro... |
2763 |
else if (!hugetlb_max_hstate) |
a34378701 hugetlb: new sysf... |
2764 2765 2766 |
mhp = &default_hstate_max_huge_pages; else mhp = &parsed_hstate->max_huge_pages; |
8faa8b077 hugetlb: support ... |
2767 |
if (mhp == last_mhp) { |
598d80914 mm: convert pr_wa... |
2768 2769 |
pr_warn("hugepages= specified twice without interleaving hugepagesz=, ignoring "); |
8faa8b077 hugetlb: support ... |
2770 2771 |
return 1; } |
a34378701 hugetlb: new sysf... |
2772 2773 |
if (sscanf(s, "%lu", mhp) <= 0) *mhp = 0; |
8faa8b077 hugetlb: support ... |
2774 2775 2776 2777 2778 |
/* * 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... |
2779 |
if (hugetlb_max_hstate && parsed_hstate->order >= MAX_ORDER) |
8faa8b077 hugetlb: support ... |
2780 2781 2782 |
hugetlb_hstate_alloc_pages(parsed_hstate); last_mhp = mhp; |
a34378701 hugetlb: new sysf... |
2783 2784 |
return 1; } |
e11bfbfcb hugetlb: override... |
2785 2786 2787 2788 2789 2790 2791 2792 |
__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... |
2793 |
|
8a2134605 hugetlb: fix CONF... |
2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 |
static unsigned int cpuset_mems_nr(unsigned int *array) { int node; unsigned int nr = 0; for_each_node_mask(node, cpuset_current_mems_allowed) nr += array[node]; return nr; } #ifdef CONFIG_SYSCTL |
06808b082 hugetlb: derive h... |
2806 2807 2808 |
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 |
2809 |
{ |
e5ff21594 hugetlb: multiple... |
2810 |
struct hstate *h = &default_hstate; |
238d3c13f mm, hugetlb: gene... |
2811 |
unsigned long tmp = h->max_huge_pages; |
08d4a2465 hugetlb: check th... |
2812 |
int ret; |
e5ff21594 hugetlb: multiple... |
2813 |
|
457c1b27e hugetlb: ensure h... |
2814 |
if (!hugepages_supported()) |
86613628b mm/hugetlb: use E... |
2815 |
return -EOPNOTSUPP; |
457c1b27e hugetlb: ensure h... |
2816 |
|
e5ff21594 hugetlb: multiple... |
2817 2818 |
table->data = &tmp; table->maxlen = sizeof(unsigned long); |
08d4a2465 hugetlb: check th... |
2819 2820 2821 |
ret = proc_doulongvec_minmax(table, write, buffer, length, ppos); if (ret) goto out; |
e5ff21594 hugetlb: multiple... |
2822 |
|
238d3c13f mm, hugetlb: gene... |
2823 2824 2825 |
if (write) ret = __nr_hugepages_store_common(obey_mempolicy, h, NUMA_NO_NODE, tmp, *length); |
08d4a2465 hugetlb: check th... |
2826 2827 |
out: return ret; |
1da177e4c Linux-2.6.12-rc2 |
2828 |
} |
396faf030 Allow huge page a... |
2829 |
|
06808b082 hugetlb: derive h... |
2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 |
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... |
2846 |
int hugetlb_overcommit_handler(struct ctl_table *table, int write, |
8d65af789 sysctl: remove "s... |
2847 |
void __user *buffer, |
a3d0c6aa1 hugetlb: add lock... |
2848 2849 |
size_t *length, loff_t *ppos) { |
a55164389 hugetlb: modular ... |
2850 |
struct hstate *h = &default_hstate; |
e5ff21594 hugetlb: multiple... |
2851 |
unsigned long tmp; |
08d4a2465 hugetlb: check th... |
2852 |
int ret; |
e5ff21594 hugetlb: multiple... |
2853 |
|
457c1b27e hugetlb: ensure h... |
2854 |
if (!hugepages_supported()) |
86613628b mm/hugetlb: use E... |
2855 |
return -EOPNOTSUPP; |
457c1b27e hugetlb: ensure h... |
2856 |
|
c033a93c0 hugetlbfs: correc... |
2857 |
tmp = h->nr_overcommit_huge_pages; |
e5ff21594 hugetlb: multiple... |
2858 |
|
bae7f4ae1 hugetlb: add hsta... |
2859 |
if (write && hstate_is_gigantic(h)) |
adbe8726d hugetlb: do not a... |
2860 |
return -EINVAL; |
e5ff21594 hugetlb: multiple... |
2861 2862 |
table->data = &tmp; table->maxlen = sizeof(unsigned long); |
08d4a2465 hugetlb: check th... |
2863 2864 2865 |
ret = proc_doulongvec_minmax(table, write, buffer, length, ppos); if (ret) goto out; |
e5ff21594 hugetlb: multiple... |
2866 2867 2868 2869 2870 2871 |
if (write) { spin_lock(&hugetlb_lock); h->nr_overcommit_huge_pages = tmp; spin_unlock(&hugetlb_lock); } |
08d4a2465 hugetlb: check th... |
2872 2873 |
out: return ret; |
a3d0c6aa1 hugetlb: add lock... |
2874 |
} |
1da177e4c Linux-2.6.12-rc2 |
2875 |
#endif /* CONFIG_SYSCTL */ |
e1759c215 proc: switch /pro... |
2876 |
void hugetlb_report_meminfo(struct seq_file *m) |
1da177e4c Linux-2.6.12-rc2 |
2877 |
{ |
a55164389 hugetlb: modular ... |
2878 |
struct hstate *h = &default_hstate; |
457c1b27e hugetlb: ensure h... |
2879 2880 |
if (!hugepages_supported()) return; |
e1759c215 proc: switch /pro... |
2881 |
seq_printf(m, |
4f98a2fee vmscan: split LRU... |
2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 |
"HugePages_Total: %5lu " "HugePages_Free: %5lu " "HugePages_Rsvd: %5lu " "HugePages_Surp: %5lu " "Hugepagesize: %8lu kB ", |
a55164389 hugetlb: modular ... |
2892 2893 2894 2895 2896 |
h->nr_huge_pages, h->free_huge_pages, h->resv_huge_pages, h->surplus_huge_pages, 1UL << (huge_page_order(h) + PAGE_SHIFT - 10)); |
1da177e4c Linux-2.6.12-rc2 |
2897 2898 2899 2900 |
} int hugetlb_report_node_meminfo(int nid, char *buf) { |
a55164389 hugetlb: modular ... |
2901 |
struct hstate *h = &default_hstate; |
457c1b27e hugetlb: ensure h... |
2902 2903 |
if (!hugepages_supported()) return 0; |
1da177e4c Linux-2.6.12-rc2 |
2904 2905 2906 |
return sprintf(buf, "Node %d HugePages_Total: %5u " |
a1de09195 hugetlb: indicate... |
2907 2908 2909 2910 |
"Node %d HugePages_Free: %5u " "Node %d HugePages_Surp: %5u ", |
a55164389 hugetlb: modular ... |
2911 2912 2913 |
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 |
2914 |
} |
949f7ec57 mm, hugetlb: incl... |
2915 2916 2917 2918 |
void hugetlb_show_meminfo(void) { struct hstate *h; int nid; |
457c1b27e hugetlb: ensure h... |
2919 2920 |
if (!hugepages_supported()) return; |
949f7ec57 mm, hugetlb: incl... |
2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 |
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... |
2931 2932 2933 2934 2935 2936 |
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 |
2937 2938 2939 |
/* Return the number pages of memory we physically have, in PAGE_SIZE units. */ unsigned long hugetlb_total_pages(void) { |
d00285884 mm/hugetlb: fix t... |
2940 2941 2942 2943 2944 2945 |
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 |
2946 |
} |
1da177e4c Linux-2.6.12-rc2 |
2947 |
|
a55164389 hugetlb: modular ... |
2948 |
static int hugetlb_acct_memory(struct hstate *h, long delta) |
fc1b8a73d hugetlb: move hug... |
2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 |
{ 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 ... |
2971 |
if (gather_surplus_pages(h, delta) < 0) |
fc1b8a73d hugetlb: move hug... |
2972 |
goto out; |
a55164389 hugetlb: modular ... |
2973 2974 |
if (delta > cpuset_mems_nr(h->free_huge_pages_node)) { return_unused_surplus_pages(h, delta); |
fc1b8a73d hugetlb: move hug... |
2975 2976 2977 2978 2979 2980 |
goto out; } } ret = 0; if (delta < 0) |
a55164389 hugetlb: modular ... |
2981 |
return_unused_surplus_pages(h, (unsigned long) -delta); |
fc1b8a73d hugetlb: move hug... |
2982 2983 2984 2985 2986 |
out: spin_unlock(&hugetlb_lock); return ret; } |
84afd99b8 hugetlb reservati... |
2987 2988 |
static void hugetlb_vm_op_open(struct vm_area_struct *vma) { |
f522c3ac0 mm, hugetlb: chan... |
2989 |
struct resv_map *resv = vma_resv_map(vma); |
84afd99b8 hugetlb reservati... |
2990 2991 2992 2993 2994 |
/* * 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... |
2995 |
* has a reference to the reservation map it cannot disappear until |
84afd99b8 hugetlb reservati... |
2996 2997 2998 |
* after this open call completes. It is therefore safe to take a * new reference here without additional locking. */ |
4e35f4838 mm, hugetlb: use ... |
2999 |
if (resv && is_vma_resv_set(vma, HPAGE_RESV_OWNER)) |
f522c3ac0 mm, hugetlb: chan... |
3000 |
kref_get(&resv->refs); |
84afd99b8 hugetlb reservati... |
3001 |
} |
a1e78772d hugetlb: reserve ... |
3002 3003 |
static void hugetlb_vm_op_close(struct vm_area_struct *vma) { |
a55164389 hugetlb: modular ... |
3004 |
struct hstate *h = hstate_vma(vma); |
f522c3ac0 mm, hugetlb: chan... |
3005 |
struct resv_map *resv = vma_resv_map(vma); |
90481622d hugepages: fix us... |
3006 |
struct hugepage_subpool *spool = subpool_vma(vma); |
4e35f4838 mm, hugetlb: use ... |
3007 |
unsigned long reserve, start, end; |
1c5ecae3a hugetlbfs: add mi... |
3008 |
long gbl_reserve; |
84afd99b8 hugetlb reservati... |
3009 |
|
4e35f4838 mm, hugetlb: use ... |
3010 3011 |
if (!resv || !is_vma_resv_set(vma, HPAGE_RESV_OWNER)) return; |
84afd99b8 hugetlb reservati... |
3012 |
|
4e35f4838 mm, hugetlb: use ... |
3013 3014 |
start = vma_hugecache_offset(h, vma, vma->vm_start); end = vma_hugecache_offset(h, vma, vma->vm_end); |
84afd99b8 hugetlb reservati... |
3015 |
|
4e35f4838 mm, hugetlb: use ... |
3016 |
reserve = (end - start) - region_count(resv, start, end); |
84afd99b8 hugetlb reservati... |
3017 |
|
4e35f4838 mm, hugetlb: use ... |
3018 3019 3020 |
kref_put(&resv->refs, resv_map_release); if (reserve) { |
1c5ecae3a hugetlbfs: add mi... |
3021 3022 3023 3024 3025 3026 |
/* * 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... |
3027 |
} |
a1e78772d hugetlb: reserve ... |
3028 |
} |
1da177e4c Linux-2.6.12-rc2 |
3029 3030 3031 3032 3033 3034 |
/* * We cannot handle pagefaults against hugetlb pages at all. They cause * handle_mm_fault() to try to instantiate regular-sized pages in the * hugegpage VMA. do_page_fault() is supposed to trap this, so BUG is we get * this far. */ |
d0217ac04 mm: fault feedbac... |
3035 |
static int hugetlb_vm_op_fault(struct vm_area_struct *vma, struct vm_fault *vmf) |
1da177e4c Linux-2.6.12-rc2 |
3036 3037 |
{ BUG(); |
d0217ac04 mm: fault feedbac... |
3038 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
3039 |
} |
f0f37e2f7 const: mark struc... |
3040 |
const struct vm_operations_struct hugetlb_vm_ops = { |
d0217ac04 mm: fault feedbac... |
3041 |
.fault = hugetlb_vm_op_fault, |
84afd99b8 hugetlb reservati... |
3042 |
.open = hugetlb_vm_op_open, |
a1e78772d hugetlb: reserve ... |
3043 |
.close = hugetlb_vm_op_close, |
1da177e4c Linux-2.6.12-rc2 |
3044 |
}; |
1e8f889b1 [PATCH] Hugetlb: ... |
3045 3046 |
static pte_t make_huge_pte(struct vm_area_struct *vma, struct page *page, int writable) |
63551ae0f [PATCH] Hugepage ... |
3047 3048 |
{ pte_t entry; |
1e8f889b1 [PATCH] Hugetlb: ... |
3049 |
if (writable) { |
106c992a5 mm/hugetlb: add m... |
3050 3051 |
entry = huge_pte_mkwrite(huge_pte_mkdirty(mk_huge_pte(page, vma->vm_page_prot))); |
63551ae0f [PATCH] Hugepage ... |
3052 |
} else { |
106c992a5 mm/hugetlb: add m... |
3053 3054 |
entry = huge_pte_wrprotect(mk_huge_pte(page, vma->vm_page_prot)); |
63551ae0f [PATCH] Hugepage ... |
3055 3056 3057 |
} entry = pte_mkyoung(entry); entry = pte_mkhuge(entry); |
d9ed9faac mm: add new arch_... |
3058 |
entry = arch_make_huge_pte(entry, vma, page, writable); |
63551ae0f [PATCH] Hugepage ... |
3059 3060 3061 |
return entry; } |
1e8f889b1 [PATCH] Hugetlb: ... |
3062 3063 3064 3065 |
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... |
3066 |
entry = huge_pte_mkwrite(huge_pte_mkdirty(huge_ptep_get(ptep))); |
32f84528f mm: hugetlb: fix ... |
3067 |
if (huge_ptep_set_access_flags(vma, address, ptep, entry, 1)) |
4b3073e1c MM: Pass a PTE po... |
3068 |
update_mmu_cache(vma, address, ptep); |
1e8f889b1 [PATCH] Hugetlb: ... |
3069 |
} |
4a705fef9 hugetlb: fix copy... |
3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 |
static int is_hugetlb_entry_migration(pte_t pte) { swp_entry_t swp; if (huge_pte_none(pte) || pte_present(pte)) return 0; swp = pte_to_swp_entry(pte); if (non_swap_entry(swp) && is_migration_entry(swp)) return 1; else return 0; } static int is_hugetlb_entry_hwpoisoned(pte_t pte) { swp_entry_t swp; if (huge_pte_none(pte) || pte_present(pte)) return 0; swp = pte_to_swp_entry(pte); if (non_swap_entry(swp) && is_hwpoison_entry(swp)) return 1; else return 0; } |
1e8f889b1 [PATCH] Hugetlb: ... |
3095 |
|
63551ae0f [PATCH] Hugepage ... |
3096 3097 3098 3099 3100 |
int copy_hugetlb_page_range(struct mm_struct *dst, struct mm_struct *src, struct vm_area_struct *vma) { pte_t *src_pte, *dst_pte, entry; struct page *ptepage; |
1c59827d1 [PATCH] mm: huget... |
3101 |
unsigned long addr; |
1e8f889b1 [PATCH] Hugetlb: ... |
3102 |
int cow; |
a55164389 hugetlb: modular ... |
3103 3104 |
struct hstate *h = hstate_vma(vma); unsigned long sz = huge_page_size(h); |
e8569dd29 mm/hugetlb.c: cal... |
3105 3106 3107 |
unsigned long mmun_start; /* For mmu_notifiers */ unsigned long mmun_end; /* For mmu_notifiers */ int ret = 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
3108 3109 |
cow = (vma->vm_flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE; |
63551ae0f [PATCH] Hugepage ... |
3110 |
|
e8569dd29 mm/hugetlb.c: cal... |
3111 3112 3113 3114 |
mmun_start = vma->vm_start; mmun_end = vma->vm_end; if (cow) mmu_notifier_invalidate_range_start(src, mmun_start, mmun_end); |
a55164389 hugetlb: modular ... |
3115 |
for (addr = vma->vm_start; addr < vma->vm_end; addr += sz) { |
cb900f412 mm, hugetlb: conv... |
3116 |
spinlock_t *src_ptl, *dst_ptl; |
c74df32c7 [PATCH] mm: ptd_a... |
3117 3118 3119 |
src_pte = huge_pte_offset(src, addr); if (!src_pte) continue; |
a55164389 hugetlb: modular ... |
3120 |
dst_pte = huge_pte_alloc(dst, addr, sz); |
e8569dd29 mm/hugetlb.c: cal... |
3121 3122 3123 3124 |
if (!dst_pte) { ret = -ENOMEM; break; } |
c5c99429f fix hugepages lea... |
3125 3126 3127 3128 |
/* If the pagetables are shared don't copy or take references */ if (dst_pte == src_pte) continue; |
cb900f412 mm, hugetlb: conv... |
3129 3130 3131 |
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... |
3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 |
entry = huge_ptep_get(src_pte); if (huge_pte_none(entry)) { /* skip none entry */ ; } else if (unlikely(is_hugetlb_entry_migration(entry) || is_hugetlb_entry_hwpoisoned(entry))) { swp_entry_t swp_entry = pte_to_swp_entry(entry); if (is_write_migration_entry(swp_entry) && cow) { /* * COW mappings require pages in both * parent and child to be set to read. */ make_migration_entry_read(&swp_entry); entry = swp_entry_to_pte(swp_entry); set_huge_pte_at(src, addr, src_pte, entry); } set_huge_pte_at(dst, addr, dst_pte, entry); } else { |
34ee645e8 mmu_notifier: cal... |
3150 |
if (cow) { |
7f2e9525b hugetlbfs: common... |
3151 |
huge_ptep_set_wrprotect(src, addr, src_pte); |
34ee645e8 mmu_notifier: cal... |
3152 3153 3154 |
mmu_notifier_invalidate_range(src, mmun_start, mmun_end); } |
0253d634e mm: hugetlb: fix ... |
3155 |
entry = huge_ptep_get(src_pte); |
1c59827d1 [PATCH] mm: huget... |
3156 3157 |
ptepage = pte_page(entry); get_page(ptepage); |
53f9263ba mm: rework mapcou... |
3158 |
page_dup_rmap(ptepage, true); |
1c59827d1 [PATCH] mm: huget... |
3159 |
set_huge_pte_at(dst, addr, dst_pte, entry); |
5d317b2b6 mm: hugetlb: proc... |
3160 |
hugetlb_count_add(pages_per_huge_page(h), dst); |
1c59827d1 [PATCH] mm: huget... |
3161 |
} |
cb900f412 mm, hugetlb: conv... |
3162 3163 |
spin_unlock(src_ptl); spin_unlock(dst_ptl); |
63551ae0f [PATCH] Hugepage ... |
3164 |
} |
63551ae0f [PATCH] Hugepage ... |
3165 |
|
e8569dd29 mm/hugetlb.c: cal... |
3166 3167 3168 3169 |
if (cow) mmu_notifier_invalidate_range_end(src, mmun_start, mmun_end); return ret; |
63551ae0f [PATCH] Hugepage ... |
3170 |
} |
24669e584 hugetlb: use mmu_... |
3171 3172 3173 |
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 ... |
3174 3175 3176 |
{ struct mm_struct *mm = vma->vm_mm; unsigned long address; |
c7546f8f0 [PATCH] Fix hugep... |
3177 |
pte_t *ptep; |
63551ae0f [PATCH] Hugepage ... |
3178 |
pte_t pte; |
cb900f412 mm, hugetlb: conv... |
3179 |
spinlock_t *ptl; |
63551ae0f [PATCH] Hugepage ... |
3180 |
struct page *page; |
a55164389 hugetlb: modular ... |
3181 3182 |
struct hstate *h = hstate_vma(vma); unsigned long sz = huge_page_size(h); |
2ec74c3ef mm: move all mmu ... |
3183 3184 |
const unsigned long mmun_start = start; /* For mmu_notifiers */ const unsigned long mmun_end = end; /* For mmu_notifiers */ |
a55164389 hugetlb: modular ... |
3185 |
|
63551ae0f [PATCH] Hugepage ... |
3186 |
WARN_ON(!is_vm_hugetlb_page(vma)); |
a55164389 hugetlb: modular ... |
3187 3188 |
BUG_ON(start & ~huge_page_mask(h)); BUG_ON(end & ~huge_page_mask(h)); |
63551ae0f [PATCH] Hugepage ... |
3189 |
|
24669e584 hugetlb: use mmu_... |
3190 |
tlb_start_vma(tlb, vma); |
2ec74c3ef mm: move all mmu ... |
3191 |
mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); |
569f48b85 mm: hugetlb: fix ... |
3192 |
address = start; |
569f48b85 mm: hugetlb: fix ... |
3193 |
for (; address < end; address += sz) { |
c7546f8f0 [PATCH] Fix hugep... |
3194 |
ptep = huge_pte_offset(mm, address); |
4c8872659 [PATCH] hugetlb: ... |
3195 |
if (!ptep) |
c7546f8f0 [PATCH] Fix hugep... |
3196 |
continue; |
cb900f412 mm, hugetlb: conv... |
3197 |
ptl = huge_pte_lock(h, mm, ptep); |
31d49da5a mm/hugetlb: simpl... |
3198 3199 3200 3201 |
if (huge_pmd_unshare(mm, &address, ptep)) { spin_unlock(ptl); continue; } |
39dde65c9 [PATCH] shared pa... |
3202 |
|
6629326b8 mm: hugetlb: clea... |
3203 |
pte = huge_ptep_get(ptep); |
31d49da5a mm/hugetlb: simpl... |
3204 3205 3206 3207 |
if (huge_pte_none(pte)) { spin_unlock(ptl); continue; } |
6629326b8 mm: hugetlb: clea... |
3208 3209 |
/* |
9fbc1f635 mm/hugetlb: add m... |
3210 3211 |
* Migrating hugepage or HWPoisoned hugepage is already * unmapped and its refcount is dropped, so just clear pte here. |
6629326b8 mm: hugetlb: clea... |
3212 |
*/ |
9fbc1f635 mm/hugetlb: add m... |
3213 |
if (unlikely(!pte_present(pte))) { |
106c992a5 mm/hugetlb: add m... |
3214 |
huge_pte_clear(mm, address, ptep); |
31d49da5a mm/hugetlb: simpl... |
3215 3216 |
spin_unlock(ptl); continue; |
8c4894c6b hwpoison, hugetlb... |
3217 |
} |
6629326b8 mm: hugetlb: clea... |
3218 3219 |
page = pte_page(pte); |
04f2cbe35 hugetlb: guarante... |
3220 3221 3222 3223 3224 3225 |
/* * 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... |
3226 3227 3228 3229 |
if (page != ref_page) { spin_unlock(ptl); continue; } |
04f2cbe35 hugetlb: guarante... |
3230 3231 3232 3233 3234 3235 3236 |
/* * 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... |
3237 |
pte = huge_ptep_get_and_clear(mm, address, ptep); |
24669e584 hugetlb: use mmu_... |
3238 |
tlb_remove_tlb_entry(tlb, ptep, address); |
106c992a5 mm/hugetlb: add m... |
3239 |
if (huge_pte_dirty(pte)) |
6649a3863 [PATCH] hugetlb: ... |
3240 |
set_page_dirty(page); |
9e81130b7 mm: hugetlb: bail... |
3241 |
|
5d317b2b6 mm: hugetlb: proc... |
3242 |
hugetlb_count_sub(pages_per_huge_page(h), mm); |
d281ee614 rmap: add argumen... |
3243 |
page_remove_rmap(page, true); |
31d49da5a mm/hugetlb: simpl... |
3244 |
|
cb900f412 mm, hugetlb: conv... |
3245 |
spin_unlock(ptl); |
e77b0852b mm/mmu_gather: tr... |
3246 |
tlb_remove_page_size(tlb, page, huge_page_size(h)); |
31d49da5a mm/hugetlb: simpl... |
3247 3248 3249 3250 3251 |
/* * Bail out after unmapping reference page if supplied */ if (ref_page) break; |
fe1668ae5 [PATCH] enforce p... |
3252 |
} |
2ec74c3ef mm: move all mmu ... |
3253 |
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
24669e584 hugetlb: use mmu_... |
3254 |
tlb_end_vma(tlb, vma); |
1da177e4c Linux-2.6.12-rc2 |
3255 |
} |
63551ae0f [PATCH] Hugepage ... |
3256 |
|
d833352a4 mm: hugetlbfs: cl... |
3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 |
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... |
3269 |
* is to clear it before releasing the i_mmap_rwsem. This works |
d833352a4 mm: hugetlbfs: cl... |
3270 |
* because in the context this is called, the VMA is about to be |
c8c06efa8 mm: convert i_mma... |
3271 |
* destroyed and the i_mmap_rwsem is held. |
d833352a4 mm: hugetlbfs: cl... |
3272 3273 3274 |
*/ vma->vm_flags &= ~VM_MAYSHARE; } |
502717f4e [PATCH] hugetlb: ... |
3275 |
void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start, |
04f2cbe35 hugetlb: guarante... |
3276 |
unsigned long end, struct page *ref_page) |
502717f4e [PATCH] hugetlb: ... |
3277 |
{ |
24669e584 hugetlb: use mmu_... |
3278 3279 3280 3281 |
struct mm_struct *mm; struct mmu_gather tlb; mm = vma->vm_mm; |
2b047252d Fix TLB gather vi... |
3282 |
tlb_gather_mmu(&tlb, mm, start, end); |
24669e584 hugetlb: use mmu_... |
3283 3284 |
__unmap_hugepage_range(&tlb, vma, start, end, ref_page); tlb_finish_mmu(&tlb, start, end); |
502717f4e [PATCH] hugetlb: ... |
3285 |
} |
04f2cbe35 hugetlb: guarante... |
3286 3287 3288 3289 3290 3291 |
/* * 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 ... |
3292 3293 |
static void unmap_ref_private(struct mm_struct *mm, struct vm_area_struct *vma, struct page *page, unsigned long address) |
04f2cbe35 hugetlb: guarante... |
3294 |
{ |
7526674de hugetlb: make unm... |
3295 |
struct hstate *h = hstate_vma(vma); |
04f2cbe35 hugetlb: guarante... |
3296 3297 |
struct vm_area_struct *iter_vma; struct address_space *mapping; |
04f2cbe35 hugetlb: guarante... |
3298 3299 3300 3301 3302 3303 |
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... |
3304 |
address = address & huge_page_mask(h); |
36e4f20af hugetlb: do not u... |
3305 3306 |
pgoff = ((address - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff; |
93c76a3d4 file_inode(f)->i_... |
3307 |
mapping = vma->vm_file->f_mapping; |
04f2cbe35 hugetlb: guarante... |
3308 |
|
4eb2b1dcd hugetlb: acquire ... |
3309 3310 3311 3312 3313 |
/* * 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... |
3314 |
i_mmap_lock_write(mapping); |
6b2dbba8b mm: replace vma p... |
3315 |
vma_interval_tree_foreach(iter_vma, &mapping->i_mmap, pgoff, pgoff) { |
04f2cbe35 hugetlb: guarante... |
3316 3317 3318 3319 3320 |
/* Do not unmap the current VMA */ if (iter_vma == vma) continue; /* |
2f84a8990 mm: hugetlbfs: sk... |
3321 3322 3323 3324 3325 3326 3327 3328 |
* 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... |
3329 3330 3331 3332 3333 3334 3335 |
* 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_... |
3336 3337 |
unmap_hugepage_range(iter_vma, address, address + huge_page_size(h), page); |
04f2cbe35 hugetlb: guarante... |
3338 |
} |
83cde9e8b mm: use new helpe... |
3339 |
i_mmap_unlock_write(mapping); |
04f2cbe35 hugetlb: guarante... |
3340 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
3341 3342 |
/* * Hugetlb_cow() should be called with page lock of the original hugepage held. |
ef009b25f hugetlb: clarify ... |
3343 3344 3345 |
* 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... |
3346 |
*/ |
1e8f889b1 [PATCH] Hugetlb: ... |
3347 |
static int hugetlb_cow(struct mm_struct *mm, struct vm_area_struct *vma, |
66c677037 mm/hugetlb.c: use... |
3348 3349 |
unsigned long address, pte_t *ptep, struct page *pagecache_page, spinlock_t *ptl) |
1e8f889b1 [PATCH] Hugetlb: ... |
3350 |
{ |
66c677037 mm/hugetlb.c: use... |
3351 |
pte_t pte; |
a55164389 hugetlb: modular ... |
3352 |
struct hstate *h = hstate_vma(vma); |
1e8f889b1 [PATCH] Hugetlb: ... |
3353 |
struct page *old_page, *new_page; |
ad4404a22 mm,hugetlb: simpl... |
3354 |
int ret = 0, outside_reserve = 0; |
2ec74c3ef mm: move all mmu ... |
3355 3356 |
unsigned long mmun_start; /* For mmu_notifiers */ unsigned long mmun_end; /* For mmu_notifiers */ |
1e8f889b1 [PATCH] Hugetlb: ... |
3357 |
|
66c677037 mm/hugetlb.c: use... |
3358 |
pte = huge_ptep_get(ptep); |
1e8f889b1 [PATCH] Hugetlb: ... |
3359 |
old_page = pte_page(pte); |
04f2cbe35 hugetlb: guarante... |
3360 |
retry_avoidcopy: |
1e8f889b1 [PATCH] Hugetlb: ... |
3361 3362 |
/* If no-one else is actually using this page, avoid the copy * and just make the page writable */ |
37a2140dc mm, hugetlb: do n... |
3363 |
if (page_mapcount(old_page) == 1 && PageAnon(old_page)) { |
5a49973d7 mm: thp: refix fa... |
3364 |
page_move_anon_rmap(old_page, vma); |
1e8f889b1 [PATCH] Hugetlb: ... |
3365 |
set_huge_ptep_writable(vma, address, ptep); |
83c54070e mm: fault feedbac... |
3366 |
return 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
3367 |
} |
04f2cbe35 hugetlb: guarante... |
3368 3369 3370 3371 3372 3373 3374 3375 3376 |
/* * 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... |
3377 |
if (is_vma_resv_set(vma, HPAGE_RESV_OWNER) && |
04f2cbe35 hugetlb: guarante... |
3378 3379 |
old_page != pagecache_page) outside_reserve = 1; |
09cbfeaf1 mm, fs: get rid o... |
3380 |
get_page(old_page); |
b76c8cfbf hugetlb: prevent ... |
3381 |
|
ad4404a22 mm,hugetlb: simpl... |
3382 3383 3384 3385 |
/* * 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... |
3386 |
spin_unlock(ptl); |
04f2cbe35 hugetlb: guarante... |
3387 |
new_page = alloc_huge_page(vma, address, outside_reserve); |
1e8f889b1 [PATCH] Hugetlb: ... |
3388 |
|
2fc39cec6 hugetlb: debit qu... |
3389 |
if (IS_ERR(new_page)) { |
04f2cbe35 hugetlb: guarante... |
3390 3391 3392 3393 3394 3395 3396 3397 |
/* * 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... |
3398 |
put_page(old_page); |
04f2cbe35 hugetlb: guarante... |
3399 |
BUG_ON(huge_pte_none(pte)); |
2f4612af4 mm,hugetlb: make ... |
3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 |
unmap_ref_private(mm, vma, old_page, address); BUG_ON(huge_pte_none(pte)); spin_lock(ptl); ptep = huge_pte_offset(mm, address & huge_page_mask(h)); if (likely(ptep && pte_same(huge_ptep_get(ptep), pte))) goto retry_avoidcopy; /* * race occurs while re-acquiring page table * lock, and our job is done. */ return 0; |
04f2cbe35 hugetlb: guarante... |
3412 |
} |
ad4404a22 mm,hugetlb: simpl... |
3413 3414 3415 |
ret = (PTR_ERR(new_page) == -ENOMEM) ? VM_FAULT_OOM : VM_FAULT_SIGBUS; goto out_release_old; |
1e8f889b1 [PATCH] Hugetlb: ... |
3416 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
3417 3418 3419 3420 |
/* * When the original hugepage is shared one, it does not have * anon_vma prepared. */ |
44e2aa937 mm/hugetlb.c: add... |
3421 |
if (unlikely(anon_vma_prepare(vma))) { |
ad4404a22 mm,hugetlb: simpl... |
3422 3423 |
ret = VM_FAULT_OOM; goto out_release_all; |
44e2aa937 mm/hugetlb.c: add... |
3424 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
3425 |
|
47ad8475c thp: clear_copy_h... |
3426 3427 |
copy_user_huge_page(new_page, old_page, address, vma, pages_per_huge_page(h)); |
0ed361dec mm: fix PageUptod... |
3428 |
__SetPageUptodate(new_page); |
bcc542223 mm: hugetlb: intr... |
3429 |
set_page_huge_active(new_page); |
1e8f889b1 [PATCH] Hugetlb: ... |
3430 |
|
2ec74c3ef mm: move all mmu ... |
3431 3432 3433 |
mmun_start = address & huge_page_mask(h); mmun_end = mmun_start + huge_page_size(h); mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); |
ad4404a22 mm,hugetlb: simpl... |
3434 |
|
b76c8cfbf hugetlb: prevent ... |
3435 |
/* |
cb900f412 mm, hugetlb: conv... |
3436 |
* Retake the page table lock to check for racing updates |
b76c8cfbf hugetlb: prevent ... |
3437 3438 |
* before the page tables are altered */ |
cb900f412 mm, hugetlb: conv... |
3439 |
spin_lock(ptl); |
a55164389 hugetlb: modular ... |
3440 |
ptep = huge_pte_offset(mm, address & huge_page_mask(h)); |
a9af0c5df mm/hugetlb.c: add... |
3441 |
if (likely(ptep && pte_same(huge_ptep_get(ptep), pte))) { |
07443a85a mm, hugetlb: retu... |
3442 |
ClearPagePrivate(new_page); |
1e8f889b1 [PATCH] Hugetlb: ... |
3443 |
/* Break COW */ |
8fe627ec5 hugetlbfs: add mi... |
3444 |
huge_ptep_clear_flush(vma, address, ptep); |
34ee645e8 mmu_notifier: cal... |
3445 |
mmu_notifier_invalidate_range(mm, mmun_start, mmun_end); |
1e8f889b1 [PATCH] Hugetlb: ... |
3446 3447 |
set_huge_pte_at(mm, address, ptep, make_huge_pte(vma, new_page, 1)); |
d281ee614 rmap: add argumen... |
3448 |
page_remove_rmap(old_page, true); |
cd67f0d2a hugetlb, rmap: us... |
3449 |
hugepage_add_new_anon_rmap(new_page, vma, address); |
1e8f889b1 [PATCH] Hugetlb: ... |
3450 3451 3452 |
/* Make the old page be freed below */ new_page = old_page; } |
cb900f412 mm, hugetlb: conv... |
3453 |
spin_unlock(ptl); |
2ec74c3ef mm: move all mmu ... |
3454 |
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
ad4404a22 mm,hugetlb: simpl... |
3455 |
out_release_all: |
96b96a96d mm/hugetlb: fix h... |
3456 |
restore_reserve_on_error(h, vma, address, new_page); |
09cbfeaf1 mm, fs: get rid o... |
3457 |
put_page(new_page); |
ad4404a22 mm,hugetlb: simpl... |
3458 |
out_release_old: |
09cbfeaf1 mm, fs: get rid o... |
3459 |
put_page(old_page); |
8312034f3 mm, hugetlb: grab... |
3460 |
|
ad4404a22 mm,hugetlb: simpl... |
3461 3462 |
spin_lock(ptl); /* Caller expects lock to be held */ return ret; |
1e8f889b1 [PATCH] Hugetlb: ... |
3463 |
} |
04f2cbe35 hugetlb: guarante... |
3464 |
/* Return the pagecache page at a given address within a VMA */ |
a55164389 hugetlb: modular ... |
3465 3466 |
static struct page *hugetlbfs_pagecache_page(struct hstate *h, struct vm_area_struct *vma, unsigned long address) |
04f2cbe35 hugetlb: guarante... |
3467 3468 |
{ struct address_space *mapping; |
e7c4b0bfd huge page private... |
3469 |
pgoff_t idx; |
04f2cbe35 hugetlb: guarante... |
3470 3471 |
mapping = vma->vm_file->f_mapping; |
a55164389 hugetlb: modular ... |
3472 |
idx = vma_hugecache_offset(h, vma, address); |
04f2cbe35 hugetlb: guarante... |
3473 3474 3475 |
return find_lock_page(mapping, idx); } |
3ae77f43b mm: hugetlbfs_pag... |
3476 3477 3478 3479 3480 |
/* * 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... |
3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 |
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... |
3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 |
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); spin_lock(&inode->i_lock); inode->i_blocks += blocks_per_huge_page(h); spin_unlock(&inode->i_lock); return 0; } |
a1ed3dda0 MM: Make needless... |
3511 |
static int hugetlb_no_page(struct mm_struct *mm, struct vm_area_struct *vma, |
8382d914e mm, hugetlb: impr... |
3512 3513 |
struct address_space *mapping, pgoff_t idx, unsigned long address, pte_t *ptep, unsigned int flags) |
ac9b9c667 [PATCH] Fix handl... |
3514 |
{ |
a55164389 hugetlb: modular ... |
3515 |
struct hstate *h = hstate_vma(vma); |
ac9b9c667 [PATCH] Fix handl... |
3516 |
int ret = VM_FAULT_SIGBUS; |
409eb8c26 mm/hugetlb.c: und... |
3517 |
int anon_rmap = 0; |
4c8872659 [PATCH] hugetlb: ... |
3518 |
unsigned long size; |
4c8872659 [PATCH] hugetlb: ... |
3519 |
struct page *page; |
1e8f889b1 [PATCH] Hugetlb: ... |
3520 |
pte_t new_pte; |
cb900f412 mm, hugetlb: conv... |
3521 |
spinlock_t *ptl; |
4c8872659 [PATCH] hugetlb: ... |
3522 |
|
04f2cbe35 hugetlb: guarante... |
3523 3524 3525 |
/* * 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... |
3526 |
* COW. Warn that such a situation has occurred as it may not be obvious |
04f2cbe35 hugetlb: guarante... |
3527 3528 |
*/ if (is_vma_resv_set(vma, HPAGE_RESV_UNMAPPED)) { |
910154d52 mm/hugetlb: huget... |
3529 3530 |
pr_warn_ratelimited("PID %d killed due to inadequate hugepage pool ", |
ffb22af5b mm/hugetlb.c: con... |
3531 |
current->pid); |
04f2cbe35 hugetlb: guarante... |
3532 3533 |
return ret; } |
4c8872659 [PATCH] hugetlb: ... |
3534 3535 3536 3537 |
/* * Use page lock to guard against racing truncation * before we get page_table_lock. */ |
6bda666a0 [PATCH] hugepages... |
3538 3539 3540 |
retry: page = find_lock_page(mapping, idx); if (!page) { |
a55164389 hugetlb: modular ... |
3541 |
size = i_size_read(mapping->host) >> huge_page_shift(h); |
ebed4bfc8 [PATCH] hugetlb: ... |
3542 3543 |
if (idx >= size) goto out; |
04f2cbe35 hugetlb: guarante... |
3544 |
page = alloc_huge_page(vma, address, 0); |
2fc39cec6 hugetlb: debit qu... |
3545 |
if (IS_ERR(page)) { |
76dcee75c hugetlb: don't us... |
3546 3547 3548 3549 3550 |
ret = PTR_ERR(page); if (ret == -ENOMEM) ret = VM_FAULT_OOM; else ret = VM_FAULT_SIGBUS; |
6bda666a0 [PATCH] hugepages... |
3551 3552 |
goto out; } |
47ad8475c thp: clear_copy_h... |
3553 |
clear_huge_page(page, address, pages_per_huge_page(h)); |
0ed361dec mm: fix PageUptod... |
3554 |
__SetPageUptodate(page); |
bcc542223 mm: hugetlb: intr... |
3555 |
set_page_huge_active(page); |
ac9b9c667 [PATCH] Fix handl... |
3556 |
|
f83a275db mm: account for M... |
3557 |
if (vma->vm_flags & VM_MAYSHARE) { |
ab76ad540 hugetlbfs: New hu... |
3558 |
int err = huge_add_to_page_cache(page, mapping, idx); |
6bda666a0 [PATCH] hugepages... |
3559 3560 |
if (err) { put_page(page); |
6bda666a0 [PATCH] hugepages... |
3561 3562 3563 3564 |
if (err == -EEXIST) goto retry; goto out; } |
23be7468e hugetlb: fix infi... |
3565 |
} else { |
6bda666a0 [PATCH] hugepages... |
3566 |
lock_page(page); |
0fe6e20b9 hugetlb, rmap: ad... |
3567 3568 3569 3570 |
if (unlikely(anon_vma_prepare(vma))) { ret = VM_FAULT_OOM; goto backout_unlocked; } |
409eb8c26 mm/hugetlb.c: und... |
3571 |
anon_rmap = 1; |
23be7468e hugetlb: fix infi... |
3572 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
3573 |
} else { |
998b4382c hugetlb: fix meta... |
3574 3575 3576 3577 3578 3579 |
/* * 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 ... |
3580 |
ret = VM_FAULT_HWPOISON | |
972dc4de1 hugetlb: add an i... |
3581 |
VM_FAULT_SET_HINDEX(hstate_index(h)); |
998b4382c hugetlb: fix meta... |
3582 3583 |
goto backout_unlocked; } |
6bda666a0 [PATCH] hugepages... |
3584 |
} |
1e8f889b1 [PATCH] Hugetlb: ... |
3585 |
|
57303d801 hugetlbfs: alloca... |
3586 3587 3588 3589 3590 3591 |
/* * 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... |
3592 |
if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) { |
2b26736c8 allocate structur... |
3593 3594 3595 3596 |
if (vma_needs_reservation(h, vma, address) < 0) { ret = VM_FAULT_OOM; goto backout_unlocked; } |
5e9113731 mm/hugetlb: add c... |
3597 |
/* Just decrements count, does not deallocate */ |
feba16e25 mm/hugetlb: add r... |
3598 |
vma_end_reservation(h, vma, address); |
5e9113731 mm/hugetlb: add c... |
3599 |
} |
57303d801 hugetlbfs: alloca... |
3600 |
|
cb900f412 mm, hugetlb: conv... |
3601 3602 |
ptl = huge_pte_lockptr(h, mm, ptep); spin_lock(ptl); |
a55164389 hugetlb: modular ... |
3603 |
size = i_size_read(mapping->host) >> huge_page_shift(h); |
4c8872659 [PATCH] hugetlb: ... |
3604 3605 |
if (idx >= size) goto backout; |
83c54070e mm: fault feedbac... |
3606 |
ret = 0; |
7f2e9525b hugetlbfs: common... |
3607 |
if (!huge_pte_none(huge_ptep_get(ptep))) |
4c8872659 [PATCH] hugetlb: ... |
3608 |
goto backout; |
07443a85a mm, hugetlb: retu... |
3609 3610 |
if (anon_rmap) { ClearPagePrivate(page); |
409eb8c26 mm/hugetlb.c: und... |
3611 |
hugepage_add_new_anon_rmap(page, vma, address); |
ac7149045 mm: fix 'ERROR: d... |
3612 |
} else |
53f9263ba mm: rework mapcou... |
3613 |
page_dup_rmap(page, true); |
1e8f889b1 [PATCH] Hugetlb: ... |
3614 3615 3616 |
new_pte = make_huge_pte(vma, page, ((vma->vm_flags & VM_WRITE) && (vma->vm_flags & VM_SHARED))); set_huge_pte_at(mm, address, ptep, new_pte); |
5d317b2b6 mm: hugetlb: proc... |
3617 |
hugetlb_count_add(pages_per_huge_page(h), mm); |
788c7df45 hugetlb: fault fl... |
3618 |
if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) { |
1e8f889b1 [PATCH] Hugetlb: ... |
3619 |
/* Optimization, do the COW without a second fault */ |
66c677037 mm/hugetlb.c: use... |
3620 |
ret = hugetlb_cow(mm, vma, address, ptep, page, ptl); |
1e8f889b1 [PATCH] Hugetlb: ... |
3621 |
} |
cb900f412 mm, hugetlb: conv... |
3622 |
spin_unlock(ptl); |
4c8872659 [PATCH] hugetlb: ... |
3623 3624 |
unlock_page(page); out: |
ac9b9c667 [PATCH] Fix handl... |
3625 |
return ret; |
4c8872659 [PATCH] hugetlb: ... |
3626 3627 |
backout: |
cb900f412 mm, hugetlb: conv... |
3628 |
spin_unlock(ptl); |
2b26736c8 allocate structur... |
3629 |
backout_unlocked: |
4c8872659 [PATCH] hugetlb: ... |
3630 |
unlock_page(page); |
96b96a96d mm/hugetlb: fix h... |
3631 |
restore_reserve_on_error(h, vma, address, page); |
4c8872659 [PATCH] hugetlb: ... |
3632 3633 |
put_page(page); goto out; |
ac9b9c667 [PATCH] Fix handl... |
3634 |
} |
8382d914e mm, hugetlb: impr... |
3635 |
#ifdef CONFIG_SMP |
c672c7f29 mm/hugetlb: expos... |
3636 |
u32 hugetlb_fault_mutex_hash(struct hstate *h, struct mm_struct *mm, |
8382d914e mm, hugetlb: impr... |
3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 |
struct vm_area_struct *vma, struct address_space *mapping, pgoff_t idx, unsigned long address) { unsigned long key[2]; u32 hash; if (vma->vm_flags & VM_SHARED) { key[0] = (unsigned long) mapping; key[1] = idx; } else { key[0] = (unsigned long) mm; key[1] = address >> huge_page_shift(h); } hash = jhash2((u32 *)&key, sizeof(key)/sizeof(u32), 0); return hash & (num_fault_mutexes - 1); } #else /* * For uniprocesor systems we always use a single mutex, so just * return 0 and avoid the hashing overhead. */ |
c672c7f29 mm/hugetlb: expos... |
3661 |
u32 hugetlb_fault_mutex_hash(struct hstate *h, struct mm_struct *mm, |
8382d914e mm, hugetlb: impr... |
3662 3663 3664 3665 3666 3667 3668 |
struct vm_area_struct *vma, struct address_space *mapping, pgoff_t idx, unsigned long address) { return 0; } #endif |
86e5216f8 [PATCH] Hugetlb: ... |
3669 |
int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
788c7df45 hugetlb: fault fl... |
3670 |
unsigned long address, unsigned int flags) |
86e5216f8 [PATCH] Hugetlb: ... |
3671 |
{ |
8382d914e mm, hugetlb: impr... |
3672 |
pte_t *ptep, entry; |
cb900f412 mm, hugetlb: conv... |
3673 |
spinlock_t *ptl; |
1e8f889b1 [PATCH] Hugetlb: ... |
3674 |
int ret; |
8382d914e mm, hugetlb: impr... |
3675 3676 |
u32 hash; pgoff_t idx; |
0fe6e20b9 hugetlb, rmap: ad... |
3677 |
struct page *page = NULL; |
57303d801 hugetlbfs: alloca... |
3678 |
struct page *pagecache_page = NULL; |
a55164389 hugetlb: modular ... |
3679 |
struct hstate *h = hstate_vma(vma); |
8382d914e mm, hugetlb: impr... |
3680 |
struct address_space *mapping; |
0f792cf94 mm/hugetlb: fix g... |
3681 |
int need_wait_lock = 0; |
86e5216f8 [PATCH] Hugetlb: ... |
3682 |
|
1e16a539a mm/hugetlb.c: fix... |
3683 |
address &= huge_page_mask(h); |
fd6a03edd HWPOISON, hugetlb... |
3684 3685 3686 |
ptep = huge_pte_offset(mm, address); if (ptep) { entry = huge_ptep_get(ptep); |
290408d4a hugetlb: hugepage... |
3687 |
if (unlikely(is_hugetlb_entry_migration(entry))) { |
cb900f412 mm, hugetlb: conv... |
3688 |
migration_entry_wait_huge(vma, mm, ptep); |
290408d4a hugetlb: hugepage... |
3689 3690 |
return 0; } else if (unlikely(is_hugetlb_entry_hwpoisoned(entry))) |
32f84528f mm: hugetlb: fix ... |
3691 |
return VM_FAULT_HWPOISON_LARGE | |
972dc4de1 hugetlb: add an i... |
3692 |
VM_FAULT_SET_HINDEX(hstate_index(h)); |
0d777df5d mm: hugetlb: call... |
3693 3694 3695 3696 |
} else { ptep = huge_pte_alloc(mm, address, huge_page_size(h)); if (!ptep) return VM_FAULT_OOM; |
fd6a03edd HWPOISON, hugetlb... |
3697 |
} |
8382d914e mm, hugetlb: impr... |
3698 3699 |
mapping = vma->vm_file->f_mapping; idx = vma_hugecache_offset(h, vma, address); |
3935baa9b [PATCH] hugepage:... |
3700 3701 3702 3703 3704 |
/* * 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. */ |
c672c7f29 mm/hugetlb: expos... |
3705 3706 |
hash = hugetlb_fault_mutex_hash(h, mm, vma, mapping, idx, address); mutex_lock(&hugetlb_fault_mutex_table[hash]); |
8382d914e mm, hugetlb: impr... |
3707 |
|
7f2e9525b hugetlbfs: common... |
3708 3709 |
entry = huge_ptep_get(ptep); if (huge_pte_none(entry)) { |
8382d914e mm, hugetlb: impr... |
3710 |
ret = hugetlb_no_page(mm, vma, mapping, idx, address, ptep, flags); |
b4d1d99fd hugetlb: handle u... |
3711 |
goto out_mutex; |
3935baa9b [PATCH] hugepage:... |
3712 |
} |
86e5216f8 [PATCH] Hugetlb: ... |
3713 |
|
83c54070e mm: fault feedbac... |
3714 |
ret = 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
3715 |
|
57303d801 hugetlbfs: alloca... |
3716 |
/* |
0f792cf94 mm/hugetlb: fix g... |
3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 |
* 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... |
3727 3728 3729 3730 3731 3732 3733 |
* 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... |
3734 |
if ((flags & FAULT_FLAG_WRITE) && !huge_pte_write(entry)) { |
2b26736c8 allocate structur... |
3735 3736 |
if (vma_needs_reservation(h, vma, address) < 0) { ret = VM_FAULT_OOM; |
b4d1d99fd hugetlb: handle u... |
3737 |
goto out_mutex; |
2b26736c8 allocate structur... |
3738 |
} |
5e9113731 mm/hugetlb: add c... |
3739 |
/* Just decrements count, does not deallocate */ |
feba16e25 mm/hugetlb: add r... |
3740 |
vma_end_reservation(h, vma, address); |
57303d801 hugetlbfs: alloca... |
3741 |
|
f83a275db mm: account for M... |
3742 |
if (!(vma->vm_flags & VM_MAYSHARE)) |
57303d801 hugetlbfs: alloca... |
3743 3744 3745 |
pagecache_page = hugetlbfs_pagecache_page(h, vma, address); } |
0f792cf94 mm/hugetlb: fix g... |
3746 3747 3748 3749 3750 |
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... |
3751 3752 3753 3754 |
/* * 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... |
3755 3756 3757 |
*/ page = pte_page(entry); if (page != pagecache_page) |
0f792cf94 mm/hugetlb: fix g... |
3758 3759 3760 3761 |
if (!trylock_page(page)) { need_wait_lock = 1; goto out_ptl; } |
b4d1d99fd hugetlb: handle u... |
3762 |
|
0f792cf94 mm/hugetlb: fix g... |
3763 |
get_page(page); |
b4d1d99fd hugetlb: handle u... |
3764 |
|
788c7df45 hugetlb: fault fl... |
3765 |
if (flags & FAULT_FLAG_WRITE) { |
106c992a5 mm/hugetlb: add m... |
3766 |
if (!huge_pte_write(entry)) { |
66c677037 mm/hugetlb.c: use... |
3767 3768 |
ret = hugetlb_cow(mm, vma, address, ptep, pagecache_page, ptl); |
0f792cf94 mm/hugetlb: fix g... |
3769 |
goto out_put_page; |
b4d1d99fd hugetlb: handle u... |
3770 |
} |
106c992a5 mm/hugetlb: add m... |
3771 |
entry = huge_pte_mkdirty(entry); |
b4d1d99fd hugetlb: handle u... |
3772 3773 |
} entry = pte_mkyoung(entry); |
788c7df45 hugetlb: fault fl... |
3774 3775 |
if (huge_ptep_set_access_flags(vma, address, ptep, entry, flags & FAULT_FLAG_WRITE)) |
4b3073e1c MM: Pass a PTE po... |
3776 |
update_mmu_cache(vma, address, ptep); |
0f792cf94 mm/hugetlb: fix g... |
3777 3778 3779 3780 |
out_put_page: if (page != pagecache_page) unlock_page(page); put_page(page); |
cb900f412 mm, hugetlb: conv... |
3781 3782 |
out_ptl: spin_unlock(ptl); |
57303d801 hugetlbfs: alloca... |
3783 3784 3785 3786 3787 |
if (pagecache_page) { unlock_page(pagecache_page); put_page(pagecache_page); } |
b4d1d99fd hugetlb: handle u... |
3788 |
out_mutex: |
c672c7f29 mm/hugetlb: expos... |
3789 |
mutex_unlock(&hugetlb_fault_mutex_table[hash]); |
0f792cf94 mm/hugetlb: fix g... |
3790 3791 3792 3793 3794 3795 3796 3797 3798 |
/* * 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: ... |
3799 |
return ret; |
86e5216f8 [PATCH] Hugetlb: ... |
3800 |
} |
28a35716d mm: use long type... |
3801 3802 3803 3804 |
long follow_hugetlb_page(struct mm_struct *mm, struct vm_area_struct *vma, struct page **pages, struct vm_area_struct **vmas, unsigned long *position, unsigned long *nr_pages, long i, unsigned int flags) |
63551ae0f [PATCH] Hugepage ... |
3805 |
{ |
d5d4b0aa4 [PATCH] optimize ... |
3806 3807 |
unsigned long pfn_offset; unsigned long vaddr = *position; |
28a35716d mm: use long type... |
3808 |
unsigned long remainder = *nr_pages; |
a55164389 hugetlb: modular ... |
3809 |
struct hstate *h = hstate_vma(vma); |
63551ae0f [PATCH] Hugepage ... |
3810 |
|
63551ae0f [PATCH] Hugepage ... |
3811 |
while (vaddr < vma->vm_end && remainder) { |
4c8872659 [PATCH] hugetlb: ... |
3812 |
pte_t *pte; |
cb900f412 mm, hugetlb: conv... |
3813 |
spinlock_t *ptl = NULL; |
2a15efc95 mm: follow_hugetl... |
3814 |
int absent; |
4c8872659 [PATCH] hugetlb: ... |
3815 |
struct page *page; |
63551ae0f [PATCH] Hugepage ... |
3816 |
|
4c8872659 [PATCH] hugetlb: ... |
3817 |
/* |
02057967b mm, hugetlb: abor... |
3818 3819 3820 3821 3822 3823 3824 3825 3826 |
* If we have a pending SIGKILL, don't keep faulting pages and * potentially allocating memory. */ if (unlikely(fatal_signal_pending(current))) { remainder = 0; break; } /* |
4c8872659 [PATCH] hugetlb: ... |
3827 |
* Some archs (sparc64, sh*) have multiple pte_ts to |
2a15efc95 mm: follow_hugetl... |
3828 |
* each hugepage. We have to make sure we get the |
4c8872659 [PATCH] hugetlb: ... |
3829 |
* first, for the page indexing below to work. |
cb900f412 mm, hugetlb: conv... |
3830 3831 |
* * Note that page table lock is not held when pte is null. |
4c8872659 [PATCH] hugetlb: ... |
3832 |
*/ |
a55164389 hugetlb: modular ... |
3833 |
pte = huge_pte_offset(mm, vaddr & huge_page_mask(h)); |
cb900f412 mm, hugetlb: conv... |
3834 3835 |
if (pte) ptl = huge_pte_lock(h, mm, pte); |
2a15efc95 mm: follow_hugetl... |
3836 3837 3838 3839 |
absent = !pte || huge_pte_none(huge_ptep_get(pte)); /* * When coredumping, it suits get_dump_page if we just return |
3ae77f43b mm: hugetlbfs_pag... |
3840 3841 3842 3843 |
* 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... |
3844 |
*/ |
3ae77f43b mm: hugetlbfs_pag... |
3845 3846 |
if (absent && (flags & FOLL_DUMP) && !hugetlbfs_pagecache_present(h, vma, vaddr)) { |
cb900f412 mm, hugetlb: conv... |
3847 3848 |
if (pte) spin_unlock(ptl); |
2a15efc95 mm: follow_hugetl... |
3849 3850 3851 |
remainder = 0; break; } |
63551ae0f [PATCH] Hugepage ... |
3852 |
|
9cc3a5bd4 hugetlbfs: add sw... |
3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 |
/* * 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... |
3864 3865 |
((flags & FOLL_WRITE) && !huge_pte_write(huge_ptep_get(pte)))) { |
4c8872659 [PATCH] hugetlb: ... |
3866 |
int ret; |
63551ae0f [PATCH] Hugepage ... |
3867 |
|
cb900f412 mm, hugetlb: conv... |
3868 3869 |
if (pte) spin_unlock(ptl); |
2a15efc95 mm: follow_hugetl... |
3870 3871 |
ret = hugetlb_fault(mm, vma, vaddr, (flags & FOLL_WRITE) ? FAULT_FLAG_WRITE : 0); |
a89182c76 Fix VM_FAULT flag... |
3872 |
if (!(ret & VM_FAULT_ERROR)) |
4c8872659 [PATCH] hugetlb: ... |
3873 |
continue; |
63551ae0f [PATCH] Hugepage ... |
3874 |
|
4c8872659 [PATCH] hugetlb: ... |
3875 |
remainder = 0; |
4c8872659 [PATCH] hugetlb: ... |
3876 3877 |
break; } |
a55164389 hugetlb: modular ... |
3878 |
pfn_offset = (vaddr & ~huge_page_mask(h)) >> PAGE_SHIFT; |
7f2e9525b hugetlbfs: common... |
3879 |
page = pte_page(huge_ptep_get(pte)); |
d5d4b0aa4 [PATCH] optimize ... |
3880 |
same_page: |
d6692183a [PATCH] fix extra... |
3881 |
if (pages) { |
2a15efc95 mm: follow_hugetl... |
3882 |
pages[i] = mem_map_offset(page, pfn_offset); |
ddc58f27f mm: drop tail pag... |
3883 |
get_page(pages[i]); |
d6692183a [PATCH] fix extra... |
3884 |
} |
63551ae0f [PATCH] Hugepage ... |
3885 3886 3887 3888 3889 |
if (vmas) vmas[i] = vma; vaddr += PAGE_SIZE; |
d5d4b0aa4 [PATCH] optimize ... |
3890 |
++pfn_offset; |
63551ae0f [PATCH] Hugepage ... |
3891 3892 |
--remainder; ++i; |
d5d4b0aa4 [PATCH] optimize ... |
3893 |
if (vaddr < vma->vm_end && remainder && |
a55164389 hugetlb: modular ... |
3894 |
pfn_offset < pages_per_huge_page(h)) { |
d5d4b0aa4 [PATCH] optimize ... |
3895 3896 3897 3898 3899 3900 |
/* * We use pfn_offset to avoid touching the pageframes * of this compound page. */ goto same_page; } |
cb900f412 mm, hugetlb: conv... |
3901 |
spin_unlock(ptl); |
63551ae0f [PATCH] Hugepage ... |
3902 |
} |
28a35716d mm: use long type... |
3903 |
*nr_pages = remainder; |
63551ae0f [PATCH] Hugepage ... |
3904 |
*position = vaddr; |
2a15efc95 mm: follow_hugetl... |
3905 |
return i ? i : -EFAULT; |
63551ae0f [PATCH] Hugepage ... |
3906 |
} |
8f860591f [PATCH] Enable mp... |
3907 |
|
5491ae7b6 powerpc/mm/hugetl... |
3908 3909 3910 3911 3912 3913 3914 |
#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... |
3915 |
unsigned long hugetlb_change_protection(struct vm_area_struct *vma, |
8f860591f [PATCH] Enable mp... |
3916 3917 3918 3919 3920 3921 |
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 ... |
3922 |
struct hstate *h = hstate_vma(vma); |
7da4d641c mm: Count the num... |
3923 |
unsigned long pages = 0; |
8f860591f [PATCH] Enable mp... |
3924 3925 3926 |
BUG_ON(address >= end); flush_cache_range(vma, address, end); |
a5338093b mm: move mmu noti... |
3927 |
mmu_notifier_invalidate_range_start(mm, start, end); |
83cde9e8b mm: use new helpe... |
3928 |
i_mmap_lock_write(vma->vm_file->f_mapping); |
a55164389 hugetlb: modular ... |
3929 |
for (; address < end; address += huge_page_size(h)) { |
cb900f412 mm, hugetlb: conv... |
3930 |
spinlock_t *ptl; |
8f860591f [PATCH] Enable mp... |
3931 3932 3933 |
ptep = huge_pte_offset(mm, address); if (!ptep) continue; |
cb900f412 mm, hugetlb: conv... |
3934 |
ptl = huge_pte_lock(h, mm, ptep); |
7da4d641c mm: Count the num... |
3935 3936 |
if (huge_pmd_unshare(mm, &address, ptep)) { pages++; |
cb900f412 mm, hugetlb: conv... |
3937 |
spin_unlock(ptl); |
39dde65c9 [PATCH] shared pa... |
3938 |
continue; |
7da4d641c mm: Count the num... |
3939 |
} |
a8bda28d8 mm/hugetlb: add m... |
3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 |
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); set_huge_pte_at(mm, address, ptep, newpte); pages++; } spin_unlock(ptl); continue; } if (!huge_pte_none(pte)) { |
8f860591f [PATCH] Enable mp... |
3960 |
pte = huge_ptep_get_and_clear(mm, address, ptep); |
106c992a5 mm/hugetlb: add m... |
3961 |
pte = pte_mkhuge(huge_pte_modify(pte, newprot)); |
be7517d6a mm/hugetlb: set P... |
3962 |
pte = arch_make_huge_pte(pte, vma, NULL, 0); |
8f860591f [PATCH] Enable mp... |
3963 |
set_huge_pte_at(mm, address, ptep, pte); |
7da4d641c mm: Count the num... |
3964 |
pages++; |
8f860591f [PATCH] Enable mp... |
3965 |
} |
cb900f412 mm, hugetlb: conv... |
3966 |
spin_unlock(ptl); |
8f860591f [PATCH] Enable mp... |
3967 |
} |
d833352a4 mm: hugetlbfs: cl... |
3968 |
/* |
c8c06efa8 mm: convert i_mma... |
3969 |
* Must flush TLB before releasing i_mmap_rwsem: x86's huge_pmd_unshare |
d833352a4 mm: hugetlbfs: cl... |
3970 |
* may have cleared our pud entry and done put_page on the page table: |
c8c06efa8 mm: convert i_mma... |
3971 |
* once we release i_mmap_rwsem, another task can do the final put_page |
d833352a4 mm: hugetlbfs: cl... |
3972 3973 |
* and that page table be reused and filled with junk. */ |
5491ae7b6 powerpc/mm/hugetl... |
3974 |
flush_hugetlb_tlb_range(vma, start, end); |
34ee645e8 mmu_notifier: cal... |
3975 |
mmu_notifier_invalidate_range(mm, start, end); |
83cde9e8b mm: use new helpe... |
3976 |
i_mmap_unlock_write(vma->vm_file->f_mapping); |
a5338093b mm: move mmu noti... |
3977 |
mmu_notifier_invalidate_range_end(mm, start, end); |
7da4d641c mm: Count the num... |
3978 3979 |
return pages << h->order; |
8f860591f [PATCH] Enable mp... |
3980 |
} |
a1e78772d hugetlb: reserve ... |
3981 3982 |
int hugetlb_reserve_pages(struct inode *inode, long from, long to, |
5a6fe1259 Do not account fo... |
3983 |
struct vm_area_struct *vma, |
ca16d140a mm: don't access ... |
3984 |
vm_flags_t vm_flags) |
e4e574b76 hugetlb: Try to g... |
3985 |
{ |
17c9d12e1 Do not account fo... |
3986 |
long ret, chg; |
a55164389 hugetlb: modular ... |
3987 |
struct hstate *h = hstate_inode(inode); |
90481622d hugepages: fix us... |
3988 |
struct hugepage_subpool *spool = subpool_inode(inode); |
9119a41e9 mm, hugetlb: unif... |
3989 |
struct resv_map *resv_map; |
1c5ecae3a hugetlbfs: add mi... |
3990 |
long gbl_reserve; |
e4e574b76 hugetlb: Try to g... |
3991 |
|
a1e78772d hugetlb: reserve ... |
3992 |
/* |
17c9d12e1 Do not account fo... |
3993 3994 |
* 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... |
3995 |
* without using reserves |
17c9d12e1 Do not account fo... |
3996 |
*/ |
ca16d140a mm: don't access ... |
3997 |
if (vm_flags & VM_NORESERVE) |
17c9d12e1 Do not account fo... |
3998 3999 4000 |
return 0; /* |
a1e78772d hugetlb: reserve ... |
4001 4002 4003 4004 4005 |
* 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... |
4006 |
if (!vma || vma->vm_flags & VM_MAYSHARE) { |
4e35f4838 mm, hugetlb: use ... |
4007 |
resv_map = inode_resv_map(inode); |
9119a41e9 mm, hugetlb: unif... |
4008 |
|
1406ec9ba mm, hugetlb: impr... |
4009 |
chg = region_chg(resv_map, from, to); |
9119a41e9 mm, hugetlb: unif... |
4010 4011 4012 |
} else { resv_map = resv_map_alloc(); |
17c9d12e1 Do not account fo... |
4013 4014 |
if (!resv_map) return -ENOMEM; |
a1e78772d hugetlb: reserve ... |
4015 |
chg = to - from; |
84afd99b8 hugetlb reservati... |
4016 |
|
17c9d12e1 Do not account fo... |
4017 4018 4019 |
set_vma_resv_map(vma, resv_map); set_vma_resv_flags(vma, HPAGE_RESV_OWNER); } |
c50ac0508 hugetlb: fix resv... |
4020 4021 4022 4023 |
if (chg < 0) { ret = chg; goto out_err; } |
8a6301127 pretend cpuset ha... |
4024 |
|
1c5ecae3a hugetlbfs: add mi... |
4025 4026 4027 4028 4029 4030 4031 |
/* * 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... |
4032 4033 4034 |
ret = -ENOSPC; goto out_err; } |
5a6fe1259 Do not account fo... |
4035 4036 |
/* |
17c9d12e1 Do not account fo... |
4037 |
* Check enough hugepages are available for the reservation. |
90481622d hugepages: fix us... |
4038 |
* Hand the pages back to the subpool if there are not |
5a6fe1259 Do not account fo... |
4039 |
*/ |
1c5ecae3a hugetlbfs: add mi... |
4040 |
ret = hugetlb_acct_memory(h, gbl_reserve); |
68842c9b9 hugetlbfs: fix qu... |
4041 |
if (ret < 0) { |
1c5ecae3a hugetlbfs: add mi... |
4042 4043 |
/* put back original number of pages, chg */ (void)hugepage_subpool_put_pages(spool, chg); |
c50ac0508 hugetlb: fix resv... |
4044 |
goto out_err; |
68842c9b9 hugetlbfs: fix qu... |
4045 |
} |
17c9d12e1 Do not account fo... |
4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 |
/* * 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... |
4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 |
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... |
4076 |
return 0; |
c50ac0508 hugetlb: fix resv... |
4077 |
out_err: |
5e9113731 mm/hugetlb: add c... |
4078 4079 |
if (!vma || vma->vm_flags & VM_MAYSHARE) region_abort(resv_map, from, to); |
f031dd274 mm, hugetlb: remo... |
4080 4081 |
if (vma && is_vma_resv_set(vma, HPAGE_RESV_OWNER)) kref_put(&resv_map->refs, resv_map_release); |
c50ac0508 hugetlb: fix resv... |
4082 |
return ret; |
a43a8c39b [PATCH] tightenin... |
4083 |
} |
b5cec28d3 hugetlbfs: trunca... |
4084 4085 |
long hugetlb_unreserve_pages(struct inode *inode, long start, long end, long freed) |
a43a8c39b [PATCH] tightenin... |
4086 |
{ |
a55164389 hugetlb: modular ... |
4087 |
struct hstate *h = hstate_inode(inode); |
4e35f4838 mm, hugetlb: use ... |
4088 |
struct resv_map *resv_map = inode_resv_map(inode); |
9119a41e9 mm, hugetlb: unif... |
4089 |
long chg = 0; |
90481622d hugepages: fix us... |
4090 |
struct hugepage_subpool *spool = subpool_inode(inode); |
1c5ecae3a hugetlbfs: add mi... |
4091 |
long gbl_reserve; |
45c682a68 hugetlb: fix i_bl... |
4092 |
|
b5cec28d3 hugetlbfs: trunca... |
4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 |
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... |
4103 |
spin_lock(&inode->i_lock); |
e4c6f8bed hugetlbfs: fix i_... |
4104 |
inode->i_blocks -= (blocks_per_huge_page(h) * freed); |
45c682a68 hugetlb: fix i_bl... |
4105 |
spin_unlock(&inode->i_lock); |
1c5ecae3a hugetlbfs: add mi... |
4106 4107 4108 4109 4110 4111 |
/* * 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... |
4112 4113 |
return 0; |
a43a8c39b [PATCH] tightenin... |
4114 |
} |
93f70f900 HWPOISON, hugetlb... |
4115 |
|
3212b535f mm: hugetlb: Copy... |
4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 |
#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_... |
4127 4128 |
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... |
4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 |
/* * 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... |
4141 |
static bool vma_shareable(struct vm_area_struct *vma, unsigned long addr) |
3212b535f mm: hugetlb: Copy... |
4142 4143 4144 4145 4146 4147 4148 4149 4150 |
{ unsigned long base = addr & PUD_MASK; unsigned long end = base + PUD_SIZE; /* * check on proper vm_flags and page table alignment */ if (vma->vm_flags & VM_MAYSHARE && vma->vm_start <= base && end <= vma->vm_end) |
31aafb45f mm/hugetlb.c: mak... |
4151 4152 |
return true; return false; |
3212b535f mm: hugetlb: Copy... |
4153 4154 4155 4156 4157 4158 4159 |
} /* * Search for a shareable pmd page for hugetlb. In any case calls pmd_alloc() * and returns the corresponding pte. While this is not necessary for the * !shared pmd case because we can allocate the pmd later as well, it makes the * code much cleaner. pmd allocation is essential for the shared case because |
c8c06efa8 mm: convert i_mma... |
4160 |
* pud has to be populated inside the same i_mmap_rwsem section - otherwise |
3212b535f mm: hugetlb: Copy... |
4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 |
* racing tasks could either miss the sharing (see huge_pte_offset) or select a * bad pmd for sharing. */ pte_t *huge_pmd_share(struct mm_struct *mm, unsigned long addr, pud_t *pud) { struct vm_area_struct *vma = find_vma(mm, addr); struct address_space *mapping = vma->vm_file->f_mapping; pgoff_t idx = ((addr - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff; struct vm_area_struct *svma; unsigned long saddr; pte_t *spte = NULL; pte_t *pte; |
cb900f412 mm, hugetlb: conv... |
4174 |
spinlock_t *ptl; |
3212b535f mm: hugetlb: Copy... |
4175 4176 4177 |
if (!vma_shareable(vma, addr)) return (pte_t *)pmd_alloc(mm, pud, addr); |
83cde9e8b mm: use new helpe... |
4178 |
i_mmap_lock_write(mapping); |
3212b535f mm: hugetlb: Copy... |
4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 |
vma_interval_tree_foreach(svma, &mapping->i_mmap, idx, idx) { if (svma == vma) continue; saddr = page_table_shareable(svma, vma, addr, idx); if (saddr) { spte = huge_pte_offset(svma->vm_mm, saddr); if (spte) { get_page(virt_to_page(spte)); break; } } } if (!spte) goto out; |
cb900f412 mm, hugetlb: conv... |
4195 4196 |
ptl = huge_pte_lockptr(hstate_vma(vma), mm, spte); spin_lock(ptl); |
dc6c9a35b mm: account pmd p... |
4197 |
if (pud_none(*pud)) { |
3212b535f mm: hugetlb: Copy... |
4198 4199 |
pud_populate(mm, pud, (pmd_t *)((unsigned long)spte & PAGE_MASK)); |
c17b1f425 hugetlb: fix nr_p... |
4200 |
mm_inc_nr_pmds(mm); |
dc6c9a35b mm: account pmd p... |
4201 |
} else { |
3212b535f mm: hugetlb: Copy... |
4202 |
put_page(virt_to_page(spte)); |
dc6c9a35b mm: account pmd p... |
4203 |
} |
cb900f412 mm, hugetlb: conv... |
4204 |
spin_unlock(ptl); |
3212b535f mm: hugetlb: Copy... |
4205 4206 |
out: pte = (pte_t *)pmd_alloc(mm, pud, addr); |
83cde9e8b mm: use new helpe... |
4207 |
i_mmap_unlock_write(mapping); |
3212b535f mm: hugetlb: Copy... |
4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 |
return pte; } /* * unmap huge page backed by shared pte. * * Hugetlb pte page is ref counted at the time of mapping. If pte is shared * indicated by page_count > 1, unmap is achieved by clearing pud and * decrementing the ref count. If count == 1, the pte page is not shared. * |
cb900f412 mm, hugetlb: conv... |
4218 |
* called with page table lock held. |
3212b535f mm: hugetlb: Copy... |
4219 4220 4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 |
* * returns: 1 successfully unmapped a shared pte page * 0 the underlying pte page is not shared, or it is the last user */ int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep) { pgd_t *pgd = pgd_offset(mm, *addr); pud_t *pud = pud_offset(pgd, *addr); BUG_ON(page_count(virt_to_page(ptep)) == 0); if (page_count(virt_to_page(ptep)) == 1) return 0; pud_clear(pud); put_page(virt_to_page(ptep)); |
dc6c9a35b mm: account pmd p... |
4234 |
mm_dec_nr_pmds(mm); |
3212b535f mm: hugetlb: Copy... |
4235 4236 4237 |
*addr = ALIGN(*addr, HPAGE_SIZE * PTRS_PER_PTE) - HPAGE_SIZE; return 1; } |
9e5fc74c3 mm: hugetlb: Copy... |
4238 4239 4240 4241 4242 4243 |
#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... |
4244 4245 4246 4247 4248 |
int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep) { return 0; } |
9e5fc74c3 mm: hugetlb: Copy... |
4249 |
#define want_pmd_share() (0) |
3212b535f mm: hugetlb: Copy... |
4250 |
#endif /* CONFIG_ARCH_WANT_HUGE_PMD_SHARE */ |
9e5fc74c3 mm: hugetlb: Copy... |
4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 |
#ifdef CONFIG_ARCH_WANT_GENERAL_HUGETLB pte_t *huge_pte_alloc(struct mm_struct *mm, unsigned long addr, unsigned long sz) { pgd_t *pgd; pud_t *pud; pte_t *pte = NULL; pgd = pgd_offset(mm, addr); pud = pud_alloc(mm, pgd, addr); if (pud) { if (sz == PUD_SIZE) { pte = (pte_t *)pud; } else { BUG_ON(sz != PMD_SIZE); if (want_pmd_share() && pud_none(*pud)) pte = huge_pmd_share(mm, addr, pud); else pte = (pte_t *)pmd_alloc(mm, pud, addr); } } |
4e666314d mm, hugetlb: fix ... |
4272 |
BUG_ON(pte && pte_present(*pte) && !pte_huge(*pte)); |
9e5fc74c3 mm: hugetlb: Copy... |
4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 |
return pte; } pte_t *huge_pte_offset(struct mm_struct *mm, unsigned long addr) { pgd_t *pgd; pud_t *pud; pmd_t *pmd = NULL; pgd = pgd_offset(mm, addr); if (pgd_present(*pgd)) { pud = pud_offset(pgd, addr); if (pud_present(*pud)) { if (pud_huge(*pud)) return (pte_t *)pud; pmd = pmd_offset(pud, addr); } } return (pte_t *) pmd; } |
61f77eda9 mm/hugetlb: reduc... |
4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 |
#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 |
9e5fc74c3 mm: hugetlb: Copy... |
4308 |
follow_huge_pmd(struct mm_struct *mm, unsigned long address, |
e66f17ff7 mm/hugetlb: take ... |
4309 |
pmd_t *pmd, int flags) |
9e5fc74c3 mm: hugetlb: Copy... |
4310 |
{ |
e66f17ff7 mm/hugetlb: take ... |
4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 |
struct page *page = NULL; spinlock_t *ptl; 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; if (pmd_present(*pmd)) { |
975341270 mm/hugetlb: use p... |
4323 |
page = pmd_page(*pmd) + ((address & ~PMD_MASK) >> PAGE_SHIFT); |
e66f17ff7 mm/hugetlb: take ... |
4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 |
if (flags & FOLL_GET) get_page(page); } else { if (is_hugetlb_entry_migration(huge_ptep_get((pte_t *)pmd))) { 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... |
4339 4340 |
return page; } |
61f77eda9 mm/hugetlb: reduc... |
4341 |
struct page * __weak |
9e5fc74c3 mm: hugetlb: Copy... |
4342 |
follow_huge_pud(struct mm_struct *mm, unsigned long address, |
e66f17ff7 mm/hugetlb: take ... |
4343 |
pud_t *pud, int flags) |
9e5fc74c3 mm: hugetlb: Copy... |
4344 |
{ |
e66f17ff7 mm/hugetlb: take ... |
4345 4346 |
if (flags & FOLL_GET) return NULL; |
9e5fc74c3 mm: hugetlb: Copy... |
4347 |
|
e66f17ff7 mm/hugetlb: take ... |
4348 |
return pte_page(*(pte_t *)pud) + ((address & ~PUD_MASK) >> PAGE_SHIFT); |
9e5fc74c3 mm: hugetlb: Copy... |
4349 |
} |
d5bd91069 hugepage: move is... |
4350 |
#ifdef CONFIG_MEMORY_FAILURE |
93f70f900 HWPOISON, hugetlb... |
4351 4352 |
/* * This function is called from memory failure code. |
93f70f900 HWPOISON, hugetlb... |
4353 |
*/ |
6de2b1aab HWPOISON, hugetlb... |
4354 |
int dequeue_hwpoisoned_huge_page(struct page *hpage) |
93f70f900 HWPOISON, hugetlb... |
4355 4356 4357 |
{ struct hstate *h = page_hstate(hpage); int nid = page_to_nid(hpage); |
6de2b1aab HWPOISON, hugetlb... |
4358 |
int ret = -EBUSY; |
93f70f900 HWPOISON, hugetlb... |
4359 4360 |
spin_lock(&hugetlb_lock); |
7e1f049ef mm: hugetlb: clea... |
4361 4362 4363 4364 4365 |
/* * Just checking !page_huge_active is not enough, because that could be * an isolated/hwpoisoned hugepage (which have >0 refcount). */ if (!page_huge_active(hpage) && !page_count(hpage)) { |
56f2fb147 mm/hugetlb.c: fix... |
4366 4367 4368 4369 4370 4371 4372 |
/* * Hwpoisoned hugepage isn't linked to activelist or freelist, * but dangling hpage->lru can trigger list-debug warnings * (this happens when we call unpoison_memory() on it), * so let it point to itself with list_del_init(). */ list_del_init(&hpage->lru); |
8c6c2ecb4 HWPOSION, hugetlb... |
4373 |
set_page_refcounted(hpage); |
6de2b1aab HWPOISON, hugetlb... |
4374 4375 4376 4377 |
h->free_huge_pages--; h->free_huge_pages_node[nid]--; ret = 0; } |
93f70f900 HWPOISON, hugetlb... |
4378 |
spin_unlock(&hugetlb_lock); |
6de2b1aab HWPOISON, hugetlb... |
4379 |
return ret; |
93f70f900 HWPOISON, hugetlb... |
4380 |
} |
6de2b1aab HWPOISON, hugetlb... |
4381 |
#endif |
31caf665e mm: migrate: make... |
4382 4383 4384 |
bool isolate_huge_page(struct page *page, struct list_head *list) { |
bcc542223 mm: hugetlb: intr... |
4385 |
bool ret = true; |
309381fea mm: dump page whe... |
4386 |
VM_BUG_ON_PAGE(!PageHead(page), page); |
31caf665e mm: migrate: make... |
4387 |
spin_lock(&hugetlb_lock); |
bcc542223 mm: hugetlb: intr... |
4388 4389 4390 4391 4392 |
if (!page_huge_active(page) || !get_page_unless_zero(page)) { ret = false; goto unlock; } clear_page_huge_active(page); |
31caf665e mm: migrate: make... |
4393 |
list_move_tail(&page->lru, list); |
bcc542223 mm: hugetlb: intr... |
4394 |
unlock: |
31caf665e mm: migrate: make... |
4395 |
spin_unlock(&hugetlb_lock); |
bcc542223 mm: hugetlb: intr... |
4396 |
return ret; |
31caf665e mm: migrate: make... |
4397 4398 4399 4400 |
} void putback_active_hugepage(struct page *page) { |
309381fea mm: dump page whe... |
4401 |
VM_BUG_ON_PAGE(!PageHead(page), page); |
31caf665e mm: migrate: make... |
4402 |
spin_lock(&hugetlb_lock); |
bcc542223 mm: hugetlb: intr... |
4403 |
set_page_huge_active(page); |
31caf665e mm: migrate: make... |
4404 4405 4406 4407 |
list_move_tail(&page->lru, &(page_hstate(page))->hugepage_activelist); spin_unlock(&hugetlb_lock); put_page(page); } |