Blame view
mm/hugetlb.c
103 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 7 8 |
#include <linux/list.h> #include <linux/init.h> #include <linux/module.h> #include <linux/mm.h> |
e1759c215 proc: switch /pro... |
9 |
#include <linux/seq_file.h> |
1da177e4c Linux-2.6.12-rc2 |
10 11 |
#include <linux/sysctl.h> #include <linux/highmem.h> |
cddb8a5c1 mmu-notifiers: core |
12 |
#include <linux/mmu_notifier.h> |
1da177e4c Linux-2.6.12-rc2 |
13 |
#include <linux/nodemask.h> |
63551ae0f [PATCH] Hugepage ... |
14 |
#include <linux/pagemap.h> |
5da7ca860 [PATCH] Add NUMA ... |
15 |
#include <linux/mempolicy.h> |
3b32123d7 mm: use macros fr... |
16 |
#include <linux/compiler.h> |
aea47ff36 [PATCH] mm: make ... |
17 |
#include <linux/cpuset.h> |
3935baa9b [PATCH] hugepage:... |
18 |
#include <linux/mutex.h> |
aa888a749 hugetlb: support ... |
19 |
#include <linux/bootmem.h> |
a34378701 hugetlb: new sysf... |
20 |
#include <linux/sysfs.h> |
5a0e3ad6a include cleanup: ... |
21 |
#include <linux/slab.h> |
0fe6e20b9 hugetlb, rmap: ad... |
22 |
#include <linux/rmap.h> |
fd6a03edd HWPOISON, hugetlb... |
23 24 |
#include <linux/swap.h> #include <linux/swapops.h> |
c8721bbbd mm: memory-hotplu... |
25 |
#include <linux/page-isolation.h> |
8382d914e mm, hugetlb: impr... |
26 |
#include <linux/jhash.h> |
d6606683a Revert duplicate ... |
27 |
|
63551ae0f [PATCH] Hugepage ... |
28 29 |
#include <asm/page.h> #include <asm/pgtable.h> |
24669e584 hugetlb: use mmu_... |
30 |
#include <asm/tlb.h> |
63551ae0f [PATCH] Hugepage ... |
31 |
|
24669e584 hugetlb: use mmu_... |
32 |
#include <linux/io.h> |
63551ae0f [PATCH] Hugepage ... |
33 |
#include <linux/hugetlb.h> |
9dd540e23 hugetlb/cgroup: a... |
34 |
#include <linux/hugetlb_cgroup.h> |
9a3052306 hugetlb: add per ... |
35 |
#include <linux/node.h> |
7835e98b2 [PATCH] remove se... |
36 |
#include "internal.h" |
1da177e4c Linux-2.6.12-rc2 |
37 |
|
753162cd8 mm: hugetlb: fix ... |
38 |
int hugepages_treat_as_movable; |
a55164389 hugetlb: modular ... |
39 |
|
c3f38a387 hugetlb: make som... |
40 |
int hugetlb_max_hstate __read_mostly; |
e5ff21594 hugetlb: multiple... |
41 42 |
unsigned int default_hstate_idx; struct hstate hstates[HUGE_MAX_HSTATE]; |
c791ad1e4 mm/hugetlb: intro... |
43 44 45 46 47 |
/* * 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... |
48 |
|
53ba51d21 hugetlb: allow ar... |
49 |
__initdata LIST_HEAD(huge_boot_pages); |
e5ff21594 hugetlb: multiple... |
50 51 52 |
/* for command line parsing */ static struct hstate * __initdata parsed_hstate; static unsigned long __initdata default_hstate_max_huge_pages; |
e11bfbfcb hugetlb: override... |
53 |
static unsigned long __initdata default_hstate_size; |
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 66 |
/* * Serializes faults on the same logical page. This is used to * prevent spurious OOMs when the hugepage pool is fully utilized. */ static int num_fault_mutexes; static struct mutex *htlb_fault_mutex_table ____cacheline_aligned_in_smp; |
7ca02d0ae hugetlbfs: accept... |
67 68 |
/* Forward declaration */ static int hugetlb_acct_memory(struct hstate *h, long delta); |
90481622d hugepages: fix us... |
69 70 71 72 73 74 75 |
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... |
76 77 78 79 80 81 |
* 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... |
82 |
kfree(spool); |
7ca02d0ae hugetlbfs: accept... |
83 |
} |
90481622d hugepages: fix us... |
84 |
} |
7ca02d0ae hugetlbfs: accept... |
85 86 |
struct hugepage_subpool *hugepage_new_subpool(struct hstate *h, long max_hpages, long min_hpages) |
90481622d hugepages: fix us... |
87 88 |
{ struct hugepage_subpool *spool; |
c6a918200 hugetlbfs: add mi... |
89 |
spool = kzalloc(sizeof(*spool), GFP_KERNEL); |
90481622d hugepages: fix us... |
90 91 92 93 94 |
if (!spool) return NULL; spin_lock_init(&spool->lock); spool->count = 1; |
7ca02d0ae hugetlbfs: accept... |
95 96 97 98 99 100 101 102 103 |
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... |
104 105 106 107 108 109 110 111 112 113 114 |
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... |
115 116 117 118 119 120 121 122 123 |
/* * 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... |
124 125 |
long delta) { |
1c5ecae3a hugetlbfs: add mi... |
126 |
long ret = delta; |
90481622d hugepages: fix us... |
127 128 |
if (!spool) |
1c5ecae3a hugetlbfs: add mi... |
129 |
return ret; |
90481622d hugepages: fix us... |
130 131 |
spin_lock(&spool->lock); |
1c5ecae3a hugetlbfs: add mi... |
132 133 134 135 136 137 138 139 |
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... |
140 |
} |
90481622d hugepages: fix us... |
141 |
|
1c5ecae3a hugetlbfs: add mi... |
142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 |
if (spool->min_hpages != -1) { /* minimum size accounting */ 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... |
158 159 |
return ret; } |
1c5ecae3a hugetlbfs: add mi... |
160 161 162 163 164 165 166 |
/* * 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... |
167 168 |
long delta) { |
1c5ecae3a hugetlbfs: add mi... |
169 |
long ret = delta; |
90481622d hugepages: fix us... |
170 |
if (!spool) |
1c5ecae3a hugetlbfs: add mi... |
171 |
return delta; |
90481622d hugepages: fix us... |
172 173 |
spin_lock(&spool->lock); |
1c5ecae3a hugetlbfs: add mi... |
174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 |
if (spool->max_hpages != -1) /* maximum size accounting */ spool->used_hpages -= delta; if (spool->min_hpages != -1) { /* minimum size accounting */ 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... |
193 |
unlock_or_release_subpool(spool); |
1c5ecae3a hugetlbfs: add mi... |
194 195 |
return ret; |
90481622d hugepages: fix us... |
196 197 198 199 200 201 202 203 204 |
} 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_... |
205 |
return subpool_inode(file_inode(vma->vm_file)); |
90481622d hugepages: fix us... |
206 |
} |
e7c4b0bfd huge page private... |
207 |
/* |
968229048 hugetlb: move res... |
208 209 |
* Region tracking -- allows tracking of reservations and instantiated pages * across the pages in a mapping. |
84afd99b8 hugetlb reservati... |
210 |
* |
7b24d8616 mm, hugetlb: fix ... |
211 212 |
* The region data structures are embedded into a resv_map and * protected by a resv_map's lock |
968229048 hugetlb: move res... |
213 214 215 216 217 218 |
*/ struct file_region { struct list_head link; long from; long to; }; |
1406ec9ba mm, hugetlb: impr... |
219 |
static long region_add(struct resv_map *resv, long f, long t) |
968229048 hugetlb: move res... |
220 |
{ |
1406ec9ba mm, hugetlb: impr... |
221 |
struct list_head *head = &resv->regions; |
968229048 hugetlb: move res... |
222 |
struct file_region *rg, *nrg, *trg; |
7b24d8616 mm, hugetlb: fix ... |
223 |
spin_lock(&resv->lock); |
968229048 hugetlb: move res... |
224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 |
/* Locate the region we are either in or before. */ list_for_each_entry(rg, head, link) if (f <= rg->to) break; /* Round our left edge to the current segment if it encloses us. */ if (f > rg->from) f = rg->from; /* Check for and consume any regions we now overlap with. */ nrg = rg; list_for_each_entry_safe(rg, trg, rg->link.prev, link) { if (&rg->link == head) break; if (rg->from > t) break; /* If this area reaches higher then extend our area to * include it completely. If this is not the first area * which we intend to reuse, free it. */ if (rg->to > t) t = rg->to; if (rg != nrg) { list_del(&rg->link); kfree(rg); } } nrg->from = f; nrg->to = t; |
7b24d8616 mm, hugetlb: fix ... |
253 |
spin_unlock(&resv->lock); |
968229048 hugetlb: move res... |
254 255 |
return 0; } |
1406ec9ba mm, hugetlb: impr... |
256 |
static long region_chg(struct resv_map *resv, long f, long t) |
968229048 hugetlb: move res... |
257 |
{ |
1406ec9ba mm, hugetlb: impr... |
258 |
struct list_head *head = &resv->regions; |
7b24d8616 mm, hugetlb: fix ... |
259 |
struct file_region *rg, *nrg = NULL; |
968229048 hugetlb: move res... |
260 |
long chg = 0; |
7b24d8616 mm, hugetlb: fix ... |
261 262 |
retry: spin_lock(&resv->lock); |
968229048 hugetlb: move res... |
263 264 265 266 267 268 269 270 271 |
/* 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 ... |
272 273 274 275 276 277 278 279 280 281 282 |
if (!nrg) { spin_unlock(&resv->lock); nrg = kmalloc(sizeof(*nrg), GFP_KERNEL); if (!nrg) return -ENOMEM; nrg->from = f; nrg->to = f; INIT_LIST_HEAD(&nrg->link); goto retry; } |
968229048 hugetlb: move res... |
283 |
|
7b24d8616 mm, hugetlb: fix ... |
284 285 286 |
list_add(&nrg->link, rg->link.prev); chg = t - f; goto out_nrg; |
968229048 hugetlb: move res... |
287 288 289 290 291 292 293 294 295 296 297 298 |
} /* 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 ... |
299 |
goto out; |
968229048 hugetlb: move res... |
300 |
|
25985edce Fix common misspe... |
301 |
/* We overlap with this area, if it extends further than |
968229048 hugetlb: move res... |
302 303 304 305 306 307 308 309 |
* 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 ... |
310 311 312 313 314 315 316 317 |
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... |
318 319 |
return chg; } |
1406ec9ba mm, hugetlb: impr... |
320 |
static long region_truncate(struct resv_map *resv, long end) |
968229048 hugetlb: move res... |
321 |
{ |
1406ec9ba mm, hugetlb: impr... |
322 |
struct list_head *head = &resv->regions; |
968229048 hugetlb: move res... |
323 324 |
struct file_region *rg, *trg; long chg = 0; |
7b24d8616 mm, hugetlb: fix ... |
325 |
spin_lock(&resv->lock); |
968229048 hugetlb: move res... |
326 327 328 329 330 |
/* Locate the region we are either in or before. */ list_for_each_entry(rg, head, link) if (end <= rg->to) break; if (&rg->link == head) |
7b24d8616 mm, hugetlb: fix ... |
331 |
goto out; |
968229048 hugetlb: move res... |
332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 |
/* If we are in the middle of a region then adjust it. */ if (end > rg->from) { chg = rg->to - end; rg->to = end; rg = list_entry(rg->link.next, typeof(*rg), link); } /* Drop any remaining regions. */ list_for_each_entry_safe(rg, trg, rg->link.prev, link) { if (&rg->link == head) break; chg += rg->to - rg->from; list_del(&rg->link); kfree(rg); } |
7b24d8616 mm, hugetlb: fix ... |
348 349 350 |
out: spin_unlock(&resv->lock); |
968229048 hugetlb: move res... |
351 352 |
return chg; } |
1406ec9ba mm, hugetlb: impr... |
353 |
static long region_count(struct resv_map *resv, long f, long t) |
84afd99b8 hugetlb reservati... |
354 |
{ |
1406ec9ba mm, hugetlb: impr... |
355 |
struct list_head *head = &resv->regions; |
84afd99b8 hugetlb reservati... |
356 357 |
struct file_region *rg; long chg = 0; |
7b24d8616 mm, hugetlb: fix ... |
358 |
spin_lock(&resv->lock); |
84afd99b8 hugetlb reservati... |
359 360 |
/* Locate each segment we overlap with, and count that overlap. */ list_for_each_entry(rg, head, link) { |
f2135a4a5 mm/hugetlb.c: use... |
361 362 |
long seg_from; long seg_to; |
84afd99b8 hugetlb reservati... |
363 364 365 366 367 368 369 370 371 372 373 |
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 ... |
374 |
spin_unlock(&resv->lock); |
84afd99b8 hugetlb reservati... |
375 376 377 |
return chg; } |
968229048 hugetlb: move res... |
378 |
/* |
e7c4b0bfd huge page private... |
379 |
* Convert the address within this vma to the page offset within |
e7c4b0bfd huge page private... |
380 381 |
* the mapping, in pagecache page units; huge pages here. */ |
a55164389 hugetlb: modular ... |
382 383 |
static pgoff_t vma_hugecache_offset(struct hstate *h, struct vm_area_struct *vma, unsigned long address) |
e7c4b0bfd huge page private... |
384 |
{ |
a55164389 hugetlb: modular ... |
385 386 |
return ((address - vma->vm_start) >> huge_page_shift(h)) + (vma->vm_pgoff >> huge_page_order(h)); |
e7c4b0bfd huge page private... |
387 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
388 389 390 391 392 |
pgoff_t linear_hugepage_index(struct vm_area_struct *vma, unsigned long address) { return vma_hugecache_offset(hstate_vma(vma), vma, address); } |
84afd99b8 hugetlb reservati... |
393 |
/* |
08fba6998 mm: report the pa... |
394 395 396 397 398 399 400 401 402 403 404 |
* 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... |
405 |
return 1UL << huge_page_shift(hstate); |
08fba6998 mm: report the pa... |
406 |
} |
f340ca0f0 hugetlbfs: export... |
407 |
EXPORT_SYMBOL_GPL(vma_kernel_pagesize); |
08fba6998 mm: report the pa... |
408 409 |
/* |
3340289dd mm: report the MM... |
410 411 412 413 414 415 416 417 418 419 420 421 422 |
* 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... |
423 424 425 426 427 428 |
* 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... |
429 |
#define HPAGE_RESV_MASK (HPAGE_RESV_OWNER | HPAGE_RESV_UNMAPPED) |
84afd99b8 hugetlb reservati... |
430 |
|
a1e78772d hugetlb: reserve ... |
431 432 433 434 435 436 437 438 439 |
/* * 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... |
440 441 442 443 444 445 446 447 448 |
* * 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 ... |
449 |
*/ |
e7c4b0bfd huge page private... |
450 451 452 453 454 455 456 457 458 459 |
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... |
460 |
struct resv_map *resv_map_alloc(void) |
84afd99b8 hugetlb reservati... |
461 462 463 464 465 466 |
{ struct resv_map *resv_map = kmalloc(sizeof(*resv_map), GFP_KERNEL); if (!resv_map) return NULL; kref_init(&resv_map->refs); |
7b24d8616 mm, hugetlb: fix ... |
467 |
spin_lock_init(&resv_map->lock); |
84afd99b8 hugetlb reservati... |
468 469 470 471 |
INIT_LIST_HEAD(&resv_map->regions); return resv_map; } |
9119a41e9 mm, hugetlb: unif... |
472 |
void resv_map_release(struct kref *ref) |
84afd99b8 hugetlb reservati... |
473 474 475 476 |
{ struct resv_map *resv_map = container_of(ref, struct resv_map, refs); /* Clear out any active regions before we release the map. */ |
1406ec9ba mm, hugetlb: impr... |
477 |
region_truncate(resv_map, 0); |
84afd99b8 hugetlb reservati... |
478 479 |
kfree(resv_map); } |
4e35f4838 mm, hugetlb: use ... |
480 481 482 483 |
static inline struct resv_map *inode_resv_map(struct inode *inode) { return inode->i_mapping->private_data; } |
84afd99b8 hugetlb reservati... |
484 |
static struct resv_map *vma_resv_map(struct vm_area_struct *vma) |
a1e78772d hugetlb: reserve ... |
485 |
{ |
81d1b09c6 mm: convert a few... |
486 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); |
4e35f4838 mm, hugetlb: use ... |
487 488 489 490 491 492 493 |
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... |
494 495 |
return (struct resv_map *)(get_vma_private_data(vma) & ~HPAGE_RESV_MASK); |
4e35f4838 mm, hugetlb: use ... |
496 |
} |
a1e78772d hugetlb: reserve ... |
497 |
} |
84afd99b8 hugetlb reservati... |
498 |
static void set_vma_resv_map(struct vm_area_struct *vma, struct resv_map *map) |
a1e78772d hugetlb: reserve ... |
499 |
{ |
81d1b09c6 mm: convert a few... |
500 501 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); VM_BUG_ON_VMA(vma->vm_flags & VM_MAYSHARE, vma); |
a1e78772d hugetlb: reserve ... |
502 |
|
84afd99b8 hugetlb reservati... |
503 504 |
set_vma_private_data(vma, (get_vma_private_data(vma) & HPAGE_RESV_MASK) | (unsigned long)map); |
04f2cbe35 hugetlb: guarante... |
505 506 507 508 |
} static void set_vma_resv_flags(struct vm_area_struct *vma, unsigned long flags) { |
81d1b09c6 mm: convert a few... |
509 510 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); VM_BUG_ON_VMA(vma->vm_flags & VM_MAYSHARE, vma); |
e7c4b0bfd huge page private... |
511 512 |
set_vma_private_data(vma, get_vma_private_data(vma) | flags); |
04f2cbe35 hugetlb: guarante... |
513 514 515 516 |
} static int is_vma_resv_set(struct vm_area_struct *vma, unsigned long flag) { |
81d1b09c6 mm: convert a few... |
517 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); |
e7c4b0bfd huge page private... |
518 519 |
return (get_vma_private_data(vma) & flag) != 0; |
a1e78772d hugetlb: reserve ... |
520 |
} |
04f2cbe35 hugetlb: guarante... |
521 |
/* Reset counters to 0 and clear all HPAGE_RESV_* flags */ |
a1e78772d hugetlb: reserve ... |
522 523 |
void reset_vma_resv_huge_pages(struct vm_area_struct *vma) { |
81d1b09c6 mm: convert a few... |
524 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); |
f83a275db mm: account for M... |
525 |
if (!(vma->vm_flags & VM_MAYSHARE)) |
a1e78772d hugetlb: reserve ... |
526 527 528 529 |
vma->vm_private_data = (void *)0; } /* Returns true if the VMA has associated reserve pages */ |
af0ed73e6 mm, hugetlb: decr... |
530 |
static int vma_has_reserves(struct vm_area_struct *vma, long chg) |
a1e78772d hugetlb: reserve ... |
531 |
{ |
af0ed73e6 mm, hugetlb: decr... |
532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 |
if (vma->vm_flags & VM_NORESERVE) { /* * This address is already reserved by other process(chg == 0), * so, we should decrement reserved count. Without decrementing, * reserve count remains after releasing inode, because this * allocated page will go into page cache and is regarded as * coming from reserved pool in releasing step. Currently, we * don't have any other solution to deal with this situation * properly, so add work-around here. */ if (vma->vm_flags & VM_MAYSHARE && chg == 0) return 1; else return 0; } |
a63884e92 mm, hugetlb: remo... |
547 548 |
/* Shared mappings always use reserves */ |
f83a275db mm: account for M... |
549 |
if (vma->vm_flags & VM_MAYSHARE) |
7f09ca51e hugetlb: fix a hu... |
550 |
return 1; |
a63884e92 mm, hugetlb: remo... |
551 552 553 554 555 |
/* * Only the process that called mmap() has reserves for * private mappings. */ |
7f09ca51e hugetlb: fix a hu... |
556 557 |
if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) return 1; |
a63884e92 mm, hugetlb: remo... |
558 |
|
7f09ca51e hugetlb: fix a hu... |
559 |
return 0; |
a1e78772d hugetlb: reserve ... |
560 |
} |
a55164389 hugetlb: modular ... |
561 |
static void enqueue_huge_page(struct hstate *h, struct page *page) |
1da177e4c Linux-2.6.12-rc2 |
562 563 |
{ int nid = page_to_nid(page); |
0edaecfab hugetlb: add a li... |
564 |
list_move(&page->lru, &h->hugepage_freelists[nid]); |
a55164389 hugetlb: modular ... |
565 566 |
h->free_huge_pages++; h->free_huge_pages_node[nid]++; |
1da177e4c Linux-2.6.12-rc2 |
567 |
} |
bf50bab2b hugetlb: add allo... |
568 569 570 |
static struct page *dequeue_huge_page_node(struct hstate *h, int nid) { struct page *page; |
c8721bbbd mm: memory-hotplu... |
571 572 573 574 575 576 577 578 |
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... |
579 |
return NULL; |
0edaecfab hugetlb: add a li... |
580 |
list_move(&page->lru, &h->hugepage_activelist); |
a9869b837 hugetlb: move ref... |
581 |
set_page_refcounted(page); |
bf50bab2b hugetlb: add allo... |
582 583 584 585 |
h->free_huge_pages--; h->free_huge_pages_node[nid]--; return page; } |
86cdb465c mm: prepare to re... |
586 587 588 |
/* Movability of hugepages depends on migration support. */ static inline gfp_t htlb_alloc_mask(struct hstate *h) { |
100873d7a hugetlb: rename h... |
589 |
if (hugepages_treat_as_movable || hugepage_migration_supported(h)) |
86cdb465c mm: prepare to re... |
590 591 592 593 |
return GFP_HIGHUSER_MOVABLE; else return GFP_HIGHUSER; } |
a55164389 hugetlb: modular ... |
594 595 |
static struct page *dequeue_huge_page_vma(struct hstate *h, struct vm_area_struct *vma, |
af0ed73e6 mm, hugetlb: decr... |
596 597 |
unsigned long address, int avoid_reserve, long chg) |
1da177e4c Linux-2.6.12-rc2 |
598 |
{ |
b1c12cbcd mm/hugetlb: fix w... |
599 |
struct page *page = NULL; |
480eccf9a Fix NUMA Memory P... |
600 |
struct mempolicy *mpol; |
19770b326 mm: filter based ... |
601 |
nodemask_t *nodemask; |
c0ff7453b cpuset,mm: fix no... |
602 |
struct zonelist *zonelist; |
dd1a239f6 mm: have zonelist... |
603 604 |
struct zone *zone; struct zoneref *z; |
cc9a6c877 cpuset: mm: reduc... |
605 |
unsigned int cpuset_mems_cookie; |
1da177e4c Linux-2.6.12-rc2 |
606 |
|
a1e78772d hugetlb: reserve ... |
607 608 609 610 611 |
/* * 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... |
612 |
if (!vma_has_reserves(vma, chg) && |
a55164389 hugetlb: modular ... |
613 |
h->free_huge_pages - h->resv_huge_pages == 0) |
c0ff7453b cpuset,mm: fix no... |
614 |
goto err; |
a1e78772d hugetlb: reserve ... |
615 |
|
04f2cbe35 hugetlb: guarante... |
616 |
/* If reserves cannot be used, ensure enough pages are in the pool */ |
a55164389 hugetlb: modular ... |
617 |
if (avoid_reserve && h->free_huge_pages - h->resv_huge_pages == 0) |
6eab04a87 treewide: remove ... |
618 |
goto err; |
04f2cbe35 hugetlb: guarante... |
619 |
|
9966c4bbb mm, hugetlb: move... |
620 |
retry_cpuset: |
d26914d11 mm: optimize put_... |
621 |
cpuset_mems_cookie = read_mems_allowed_begin(); |
9966c4bbb mm, hugetlb: move... |
622 |
zonelist = huge_zonelist(vma, address, |
86cdb465c mm: prepare to re... |
623 |
htlb_alloc_mask(h), &mpol, &nodemask); |
9966c4bbb mm, hugetlb: move... |
624 |
|
19770b326 mm: filter based ... |
625 626 |
for_each_zone_zonelist_nodemask(zone, z, zonelist, MAX_NR_ZONES - 1, nodemask) { |
344736f29 cpuset: simplify ... |
627 |
if (cpuset_zone_allowed(zone, htlb_alloc_mask(h))) { |
bf50bab2b hugetlb: add allo... |
628 629 |
page = dequeue_huge_page_node(h, zone_to_nid(zone)); if (page) { |
af0ed73e6 mm, hugetlb: decr... |
630 631 632 633 |
if (avoid_reserve) break; if (!vma_has_reserves(vma, chg)) break; |
07443a85a mm, hugetlb: retu... |
634 |
SetPagePrivate(page); |
af0ed73e6 mm, hugetlb: decr... |
635 |
h->resv_huge_pages--; |
bf50bab2b hugetlb: add allo... |
636 637 |
break; } |
3abf7afd4 dequeue_huge_page... |
638 |
} |
1da177e4c Linux-2.6.12-rc2 |
639 |
} |
cc9a6c877 cpuset: mm: reduc... |
640 |
|
52cd3b074 mempolicy: rework... |
641 |
mpol_cond_put(mpol); |
d26914d11 mm: optimize put_... |
642 |
if (unlikely(!page && read_mems_allowed_retry(cpuset_mems_cookie))) |
cc9a6c877 cpuset: mm: reduc... |
643 |
goto retry_cpuset; |
1da177e4c Linux-2.6.12-rc2 |
644 |
return page; |
cc9a6c877 cpuset: mm: reduc... |
645 646 |
err: |
cc9a6c877 cpuset: mm: reduc... |
647 |
return NULL; |
1da177e4c Linux-2.6.12-rc2 |
648 |
} |
1cac6f2c0 hugetlb: move hel... |
649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 |
/* * common helper functions for hstate_next_node_to_{alloc|free}. * We may have allocated or freed a huge page based on a different * nodes_allowed previously, so h->next_node_to_{alloc|free} might * be outside of *nodes_allowed. Ensure that we use an allowed * node for alloc or free. */ static int next_node_allowed(int nid, nodemask_t *nodes_allowed) { nid = next_node(nid, *nodes_allowed); if (nid == MAX_NUMNODES) nid = first_node(*nodes_allowed); VM_BUG_ON(nid >= MAX_NUMNODES); return nid; } static int get_valid_node_allowed(int nid, nodemask_t *nodes_allowed) { if (!node_isset(nid, *nodes_allowed)) nid = next_node_allowed(nid, nodes_allowed); return nid; } /* * returns the previously saved node ["this node"] from which to * allocate a persistent huge page for the pool and advance the * next node from which to allocate, handling wrap at end of node * mask. */ static int hstate_next_node_to_alloc(struct hstate *h, nodemask_t *nodes_allowed) { int nid; VM_BUG_ON(!nodes_allowed); nid = get_valid_node_allowed(h->next_nid_to_alloc, nodes_allowed); h->next_nid_to_alloc = next_node_allowed(nid, nodes_allowed); return nid; } /* * helper for free_pool_huge_page() - return the previously saved * node ["this node"] from which to free a huge page. Advance the * next node id whether or not we find a free huge page to free so * that the next attempt to free addresses the next node. */ static int hstate_next_node_to_free(struct hstate *h, nodemask_t *nodes_allowed) { int nid; VM_BUG_ON(!nodes_allowed); nid = get_valid_node_allowed(h->next_nid_to_free, nodes_allowed); h->next_nid_to_free = next_node_allowed(nid, nodes_allowed); return nid; } #define for_each_node_mask_to_alloc(hs, nr_nodes, node, mask) \ for (nr_nodes = nodes_weight(*mask); \ nr_nodes > 0 && \ ((node = hstate_next_node_to_alloc(hs, mask)) || 1); \ nr_nodes--) #define for_each_node_mask_to_free(hs, nr_nodes, node, mask) \ for (nr_nodes = nodes_weight(*mask); \ nr_nodes > 0 && \ ((node = hstate_next_node_to_free(hs, mask)) || 1); \ nr_nodes--) |
944d9fec8 hugetlb: add supp... |
721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 |
#if defined(CONFIG_CMA) && defined(CONFIG_X86_64) static void destroy_compound_gigantic_page(struct page *page, unsigned long order) { int i; int nr_pages = 1 << order; struct page *p = page + 1; for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) { __ClearPageTail(p); set_page_refcounted(p); p->first_page = NULL; } set_compound_order(page, 0); __ClearPageHead(page); } static void free_gigantic_page(struct page *page, unsigned order) { free_contig_range(page_to_pfn(page), 1 << order); } static int __alloc_gigantic_page(unsigned long start_pfn, unsigned long nr_pages) { unsigned long end_pfn = start_pfn + nr_pages; return alloc_contig_range(start_pfn, end_pfn, MIGRATE_MOVABLE); } static bool pfn_range_valid_gigantic(unsigned long start_pfn, unsigned long nr_pages) { unsigned long i, end_pfn = start_pfn + nr_pages; struct page *page; for (i = start_pfn; i < end_pfn; i++) { if (!pfn_valid(i)) return false; page = pfn_to_page(i); if (PageReserved(page)) return false; if (page_count(page) > 0) return false; if (PageHuge(page)) return false; } return true; } static bool zone_spans_last_pfn(const struct zone *zone, unsigned long start_pfn, unsigned long nr_pages) { unsigned long last_pfn = start_pfn + nr_pages - 1; return zone_spans_pfn(zone, last_pfn); } static struct page *alloc_gigantic_page(int nid, unsigned order) { unsigned long nr_pages = 1 << order; unsigned long ret, pfn, flags; struct zone *z; z = NODE_DATA(nid)->node_zones; for (; z - NODE_DATA(nid)->node_zones < MAX_NR_ZONES; z++) { spin_lock_irqsave(&z->lock, flags); pfn = ALIGN(z->zone_start_pfn, nr_pages); while (zone_spans_last_pfn(z, pfn, nr_pages)) { if (pfn_range_valid_gigantic(pfn, nr_pages)) { /* * We release the zone lock here because * alloc_contig_range() will also lock the zone * at some point. If there's an allocation * spinning on this lock, it may win the race * and cause alloc_contig_range() to fail... */ spin_unlock_irqrestore(&z->lock, flags); ret = __alloc_gigantic_page(pfn, nr_pages); if (!ret) return pfn_to_page(pfn); spin_lock_irqsave(&z->lock, flags); } pfn += nr_pages; } spin_unlock_irqrestore(&z->lock, flags); } return NULL; } static void prep_new_huge_page(struct hstate *h, struct page *page, int nid); static void prep_compound_gigantic_page(struct page *page, unsigned long order); static struct page *alloc_fresh_gigantic_page_node(struct hstate *h, int nid) { struct page *page; page = alloc_gigantic_page(nid, huge_page_order(h)); if (page) { prep_compound_gigantic_page(page, huge_page_order(h)); prep_new_huge_page(h, page, nid); } return page; } static int alloc_fresh_gigantic_page(struct hstate *h, nodemask_t *nodes_allowed) { struct page *page = NULL; int nr_nodes, node; for_each_node_mask_to_alloc(h, nr_nodes, node, nodes_allowed) { page = alloc_fresh_gigantic_page_node(h, node); if (page) return 1; } return 0; } static inline bool gigantic_page_supported(void) { return true; } #else static inline bool gigantic_page_supported(void) { return false; } static inline void free_gigantic_page(struct page *page, unsigned order) { } static inline void destroy_compound_gigantic_page(struct page *page, unsigned long order) { } static inline int alloc_fresh_gigantic_page(struct hstate *h, nodemask_t *nodes_allowed) { return 0; } #endif |
a55164389 hugetlb: modular ... |
858 |
static void update_and_free_page(struct hstate *h, struct page *page) |
6af2acb66 hugetlb: Move upd... |
859 860 |
{ int i; |
a55164389 hugetlb: modular ... |
861 |
|
944d9fec8 hugetlb: add supp... |
862 863 |
if (hstate_is_gigantic(h) && !gigantic_page_supported()) return; |
18229df5b hugetlb: pull gig... |
864 |
|
a55164389 hugetlb: modular ... |
865 866 867 |
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 ... |
868 869 |
page[i].flags &= ~(1 << PG_locked | 1 << PG_error | 1 << PG_referenced | 1 << PG_dirty | |
a7407a27c hugetlb: update_a... |
870 871 |
1 << PG_active | 1 << PG_private | 1 << PG_writeback); |
6af2acb66 hugetlb: Move upd... |
872 |
} |
309381fea mm: dump page whe... |
873 |
VM_BUG_ON_PAGE(hugetlb_cgroup_from_page(page), page); |
6af2acb66 hugetlb: Move upd... |
874 875 |
set_compound_page_dtor(page, NULL); set_page_refcounted(page); |
944d9fec8 hugetlb: add supp... |
876 877 878 879 880 881 882 |
if (hstate_is_gigantic(h)) { destroy_compound_gigantic_page(page, huge_page_order(h)); free_gigantic_page(page, huge_page_order(h)); } else { arch_release_hugepage(page); __free_pages(page, huge_page_order(h)); } |
6af2acb66 hugetlb: Move upd... |
883 |
} |
e5ff21594 hugetlb: multiple... |
884 885 886 887 888 889 890 891 892 893 |
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... |
894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 |
/* * 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... |
918 |
void free_huge_page(struct page *page) |
27a85ef1b [PATCH] hugepage:... |
919 |
{ |
a55164389 hugetlb: modular ... |
920 921 922 923 |
/* * Can't pass hstate in here because it is called from the * compound page destructor. */ |
e5ff21594 hugetlb: multiple... |
924 |
struct hstate *h = page_hstate(page); |
7893d1d50 hugetlb: Try to g... |
925 |
int nid = page_to_nid(page); |
90481622d hugepages: fix us... |
926 927 |
struct hugepage_subpool *spool = (struct hugepage_subpool *)page_private(page); |
07443a85a mm, hugetlb: retu... |
928 |
bool restore_reserve; |
27a85ef1b [PATCH] hugepage:... |
929 |
|
e5df70ab1 hugetlb: ensure w... |
930 |
set_page_private(page, 0); |
23be7468e hugetlb: fix infi... |
931 |
page->mapping = NULL; |
7893d1d50 hugetlb: Try to g... |
932 |
BUG_ON(page_count(page)); |
0fe6e20b9 hugetlb, rmap: ad... |
933 |
BUG_ON(page_mapcount(page)); |
07443a85a mm, hugetlb: retu... |
934 |
restore_reserve = PagePrivate(page); |
16c794b4f mm/hugetlb.c: cor... |
935 |
ClearPagePrivate(page); |
27a85ef1b [PATCH] hugepage:... |
936 |
|
1c5ecae3a hugetlbfs: add mi... |
937 938 939 940 941 942 943 |
/* * 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:... |
944 |
spin_lock(&hugetlb_lock); |
bcc542223 mm: hugetlb: intr... |
945 |
clear_page_huge_active(page); |
6d76dcf40 hugetlb/cgroup: a... |
946 947 |
hugetlb_cgroup_uncharge_page(hstate_index(h), pages_per_huge_page(h), page); |
07443a85a mm, hugetlb: retu... |
948 949 |
if (restore_reserve) h->resv_huge_pages++; |
944d9fec8 hugetlb: add supp... |
950 |
if (h->surplus_huge_pages_node[nid]) { |
0edaecfab hugetlb: add a li... |
951 952 |
/* remove the page from active list */ list_del(&page->lru); |
a55164389 hugetlb: modular ... |
953 954 955 |
update_and_free_page(h, page); h->surplus_huge_pages--; h->surplus_huge_pages_node[nid]--; |
7893d1d50 hugetlb: Try to g... |
956 |
} else { |
5d3a551c2 mm: hugetlb: add ... |
957 |
arch_clear_hugepage_flags(page); |
a55164389 hugetlb: modular ... |
958 |
enqueue_huge_page(h, page); |
7893d1d50 hugetlb: Try to g... |
959 |
} |
27a85ef1b [PATCH] hugepage:... |
960 961 |
spin_unlock(&hugetlb_lock); } |
a55164389 hugetlb: modular ... |
962 |
static void prep_new_huge_page(struct hstate *h, struct page *page, int nid) |
b7ba30c67 hugetlb: factor o... |
963 |
{ |
0edaecfab hugetlb: add a li... |
964 |
INIT_LIST_HEAD(&page->lru); |
b7ba30c67 hugetlb: factor o... |
965 966 |
set_compound_page_dtor(page, free_huge_page); spin_lock(&hugetlb_lock); |
9dd540e23 hugetlb/cgroup: a... |
967 |
set_hugetlb_cgroup(page, NULL); |
a55164389 hugetlb: modular ... |
968 969 |
h->nr_huge_pages++; h->nr_huge_pages_node[nid]++; |
b7ba30c67 hugetlb: factor o... |
970 971 972 |
spin_unlock(&hugetlb_lock); put_page(page); /* free it into the hugepage allocator */ } |
2906dd528 hugetlb: prep_com... |
973 |
static void prep_compound_gigantic_page(struct page *page, unsigned long order) |
20a0307c0 mm: introduce Pag... |
974 975 976 977 978 979 980 981 |
{ int i; int nr_pages = 1 << order; struct page *p = page + 1; /* we rely on prep_new_huge_page to set the destructor */ set_compound_order(page, order); __SetPageHead(page); |
ef5a22be2 mm: hugetlb: init... |
982 |
__ClearPageReserved(page); |
20a0307c0 mm: introduce Pag... |
983 |
for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) { |
ef5a22be2 mm: hugetlb: init... |
984 985 986 987 988 989 990 991 992 993 994 995 996 |
/* * 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... |
997 |
set_page_count(p, 0); |
20a0307c0 mm: introduce Pag... |
998 |
p->first_page = page; |
44fc80573 mm, hugetlb: clos... |
999 1000 1001 |
/* Make sure p->first_page is always valid for PageTail() */ smp_wmb(); __SetPageTail(p); |
20a0307c0 mm: introduce Pag... |
1002 1003 |
} } |
7795912c2 mm: document Page... |
1004 1005 1006 1007 1008 |
/* * 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... |
1009 1010 |
int PageHuge(struct page *page) { |
20a0307c0 mm: introduce Pag... |
1011 1012 1013 1014 |
if (!PageCompound(page)) return 0; page = compound_head(page); |
758f66a29 mm/hugetlb.c: sim... |
1015 |
return get_compound_page_dtor(page) == free_huge_page; |
20a0307c0 mm: introduce Pag... |
1016 |
} |
43131e141 HWPOISON, hugetlb... |
1017 |
EXPORT_SYMBOL_GPL(PageHuge); |
27c73ae75 mm: hugetlbfs: fi... |
1018 1019 1020 1021 1022 1023 |
/* * 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... |
1024 1025 |
if (!PageHead(page_head)) return 0; |
758f66a29 mm/hugetlb.c: sim... |
1026 |
return get_compound_page_dtor(page_head) == free_huge_page; |
27c73ae75 mm: hugetlbfs: fi... |
1027 |
} |
27c73ae75 mm: hugetlbfs: fi... |
1028 |
|
13d60f4b6 futex: Take hugep... |
1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 |
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 ... |
1045 |
static struct page *alloc_fresh_huge_page_node(struct hstate *h, int nid) |
1da177e4c Linux-2.6.12-rc2 |
1046 |
{ |
1da177e4c Linux-2.6.12-rc2 |
1047 |
struct page *page; |
f96efd585 hugetlb: fix race... |
1048 |
|
6484eb3e2 page allocator: d... |
1049 |
page = alloc_pages_exact_node(nid, |
86cdb465c mm: prepare to re... |
1050 |
htlb_alloc_mask(h)|__GFP_COMP|__GFP_THISNODE| |
551883ae8 page allocator: e... |
1051 |
__GFP_REPEAT|__GFP_NOWARN, |
a55164389 hugetlb: modular ... |
1052 |
huge_page_order(h)); |
1da177e4c Linux-2.6.12-rc2 |
1053 |
if (page) { |
7f2e9525b hugetlbfs: common... |
1054 |
if (arch_prepare_hugepage(page)) { |
caff3a2c3 hugetlb: call arc... |
1055 |
__free_pages(page, huge_page_order(h)); |
7b8ee84d8 mm: fix integer a... |
1056 |
return NULL; |
7f2e9525b hugetlbfs: common... |
1057 |
} |
a55164389 hugetlb: modular ... |
1058 |
prep_new_huge_page(h, page, nid); |
1da177e4c Linux-2.6.12-rc2 |
1059 |
} |
63b4613c3 hugetlb: fix huge... |
1060 1061 1062 |
return page; } |
b22610268 mm, hugetlb: fix ... |
1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 |
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 ... |
1084 1085 1086 1087 1088 1089 |
/* * 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... |
1090 1091 |
static int free_pool_huge_page(struct hstate *h, nodemask_t *nodes_allowed, bool acct_surplus) |
e8c5c8249 hugetlb: balance ... |
1092 |
{ |
b22610268 mm, hugetlb: fix ... |
1093 |
int nr_nodes, node; |
e8c5c8249 hugetlb: balance ... |
1094 |
int ret = 0; |
b22610268 mm, hugetlb: fix ... |
1095 |
for_each_node_mask_to_free(h, nr_nodes, node, nodes_allowed) { |
685f34570 hugetlb: use free... |
1096 1097 1098 1099 |
/* * If we're returning unused surplus pages, only examine * nodes with surplus pages. */ |
b22610268 mm, hugetlb: fix ... |
1100 1101 |
if ((!acct_surplus || h->surplus_huge_pages_node[node]) && !list_empty(&h->hugepage_freelists[node])) { |
e8c5c8249 hugetlb: balance ... |
1102 |
struct page *page = |
b22610268 mm, hugetlb: fix ... |
1103 |
list_entry(h->hugepage_freelists[node].next, |
e8c5c8249 hugetlb: balance ... |
1104 1105 1106 |
struct page, lru); list_del(&page->lru); h->free_huge_pages--; |
b22610268 mm, hugetlb: fix ... |
1107 |
h->free_huge_pages_node[node]--; |
685f34570 hugetlb: use free... |
1108 1109 |
if (acct_surplus) { h->surplus_huge_pages--; |
b22610268 mm, hugetlb: fix ... |
1110 |
h->surplus_huge_pages_node[node]--; |
685f34570 hugetlb: use free... |
1111 |
} |
e8c5c8249 hugetlb: balance ... |
1112 1113 |
update_and_free_page(h, page); ret = 1; |
9a76db099 hugetlb: rework h... |
1114 |
break; |
e8c5c8249 hugetlb: balance ... |
1115 |
} |
b22610268 mm, hugetlb: fix ... |
1116 |
} |
e8c5c8249 hugetlb: balance ... |
1117 1118 1119 |
return ret; } |
c8721bbbd mm: memory-hotplu... |
1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 |
/* * Dissolve a given free hugepage into free buddy pages. This function does * nothing for in-use (including surplus) hugepages. */ static void dissolve_free_huge_page(struct page *page) { spin_lock(&hugetlb_lock); if (PageHuge(page) && !page_count(page)) { struct hstate *h = page_hstate(page); int nid = page_to_nid(page); list_del(&page->lru); h->free_huge_pages--; h->free_huge_pages_node[nid]--; update_and_free_page(h, page); } spin_unlock(&hugetlb_lock); } /* * Dissolve free hugepages in a given pfn range. Used by memory hotplug to * make specified memory blocks removable from the system. * Note that start_pfn should aligned with (minimum) hugepage size. */ void dissolve_free_huge_pages(unsigned long start_pfn, unsigned long end_pfn) { |
c8721bbbd mm: memory-hotplu... |
1145 |
unsigned long pfn; |
c8721bbbd mm: memory-hotplu... |
1146 |
|
d01776393 mm: fix potential... |
1147 1148 |
if (!hugepages_supported()) return; |
c791ad1e4 mm/hugetlb: intro... |
1149 1150 |
VM_BUG_ON(!IS_ALIGNED(start_pfn, 1 << minimum_order)); for (pfn = start_pfn; pfn < end_pfn; pfn += 1 << minimum_order) |
c8721bbbd mm: memory-hotplu... |
1151 1152 |
dissolve_free_huge_page(pfn_to_page(pfn)); } |
bf50bab2b hugetlb: add allo... |
1153 |
static struct page *alloc_buddy_huge_page(struct hstate *h, int nid) |
7893d1d50 hugetlb: Try to g... |
1154 1155 |
{ struct page *page; |
bf50bab2b hugetlb: add allo... |
1156 |
unsigned int r_nid; |
7893d1d50 hugetlb: Try to g... |
1157 |
|
bae7f4ae1 hugetlb: add hsta... |
1158 |
if (hstate_is_gigantic(h)) |
aa888a749 hugetlb: support ... |
1159 |
return NULL; |
d1c3fb1f8 hugetlb: introduc... |
1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 |
/* * 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 ... |
1184 |
if (h->surplus_huge_pages >= h->nr_overcommit_huge_pages) { |
d1c3fb1f8 hugetlb: introduc... |
1185 1186 1187 |
spin_unlock(&hugetlb_lock); return NULL; } else { |
a55164389 hugetlb: modular ... |
1188 1189 |
h->nr_huge_pages++; h->surplus_huge_pages++; |
d1c3fb1f8 hugetlb: introduc... |
1190 1191 |
} spin_unlock(&hugetlb_lock); |
bf50bab2b hugetlb: add allo... |
1192 |
if (nid == NUMA_NO_NODE) |
86cdb465c mm: prepare to re... |
1193 |
page = alloc_pages(htlb_alloc_mask(h)|__GFP_COMP| |
bf50bab2b hugetlb: add allo... |
1194 1195 1196 1197 |
__GFP_REPEAT|__GFP_NOWARN, huge_page_order(h)); else page = alloc_pages_exact_node(nid, |
86cdb465c mm: prepare to re... |
1198 |
htlb_alloc_mask(h)|__GFP_COMP|__GFP_THISNODE| |
bf50bab2b hugetlb: add allo... |
1199 |
__GFP_REPEAT|__GFP_NOWARN, huge_page_order(h)); |
d1c3fb1f8 hugetlb: introduc... |
1200 |
|
caff3a2c3 hugetlb: call arc... |
1201 1202 |
if (page && arch_prepare_hugepage(page)) { __free_pages(page, huge_page_order(h)); |
ea5768c74 mm/hugetlb.c: avo... |
1203 |
page = NULL; |
caff3a2c3 hugetlb: call arc... |
1204 |
} |
d1c3fb1f8 hugetlb: introduc... |
1205 |
spin_lock(&hugetlb_lock); |
7893d1d50 hugetlb: Try to g... |
1206 |
if (page) { |
0edaecfab hugetlb: add a li... |
1207 |
INIT_LIST_HEAD(&page->lru); |
bf50bab2b hugetlb: add allo... |
1208 |
r_nid = page_to_nid(page); |
7893d1d50 hugetlb: Try to g... |
1209 |
set_compound_page_dtor(page, free_huge_page); |
9dd540e23 hugetlb/cgroup: a... |
1210 |
set_hugetlb_cgroup(page, NULL); |
d1c3fb1f8 hugetlb: introduc... |
1211 1212 1213 |
/* * We incremented the global counters already */ |
bf50bab2b hugetlb: add allo... |
1214 1215 |
h->nr_huge_pages_node[r_nid]++; h->surplus_huge_pages_node[r_nid]++; |
3b1163006 Subject: [PATCH] ... |
1216 |
__count_vm_event(HTLB_BUDDY_PGALLOC); |
d1c3fb1f8 hugetlb: introduc... |
1217 |
} else { |
a55164389 hugetlb: modular ... |
1218 1219 |
h->nr_huge_pages--; h->surplus_huge_pages--; |
3b1163006 Subject: [PATCH] ... |
1220 |
__count_vm_event(HTLB_BUDDY_PGALLOC_FAIL); |
7893d1d50 hugetlb: Try to g... |
1221 |
} |
d1c3fb1f8 hugetlb: introduc... |
1222 |
spin_unlock(&hugetlb_lock); |
7893d1d50 hugetlb: Try to g... |
1223 1224 1225 |
return page; } |
e4e574b76 hugetlb: Try to g... |
1226 |
/* |
bf50bab2b hugetlb: add allo... |
1227 1228 1229 1230 1231 1232 |
* 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... |
1233 |
struct page *page = NULL; |
bf50bab2b hugetlb: add allo... |
1234 1235 |
spin_lock(&hugetlb_lock); |
4ef918480 mm, hugetlb: prot... |
1236 1237 |
if (h->free_huge_pages - h->resv_huge_pages > 0) page = dequeue_huge_page_node(h, nid); |
bf50bab2b hugetlb: add allo... |
1238 |
spin_unlock(&hugetlb_lock); |
94ae8ba71 hugetlb/cgroup: a... |
1239 |
if (!page) |
bf50bab2b hugetlb: add allo... |
1240 1241 1242 1243 1244 1245 |
page = alloc_buddy_huge_page(h, nid); return page; } /* |
25985edce Fix common misspe... |
1246 |
* Increase the hugetlb pool such that it can accommodate a reservation |
e4e574b76 hugetlb: Try to g... |
1247 1248 |
* of size 'delta'. */ |
a55164389 hugetlb: modular ... |
1249 |
static int gather_surplus_pages(struct hstate *h, int delta) |
e4e574b76 hugetlb: Try to g... |
1250 1251 1252 1253 1254 |
{ struct list_head surplus_list; struct page *page, *tmp; int ret, i; int needed, allocated; |
28073b02b mm: hugetlb: defe... |
1255 |
bool alloc_ok = true; |
e4e574b76 hugetlb: Try to g... |
1256 |
|
a55164389 hugetlb: modular ... |
1257 |
needed = (h->resv_huge_pages + delta) - h->free_huge_pages; |
ac09b3a15 hugetlb: close a ... |
1258 |
if (needed <= 0) { |
a55164389 hugetlb: modular ... |
1259 |
h->resv_huge_pages += delta; |
e4e574b76 hugetlb: Try to g... |
1260 |
return 0; |
ac09b3a15 hugetlb: close a ... |
1261 |
} |
e4e574b76 hugetlb: Try to g... |
1262 1263 1264 1265 1266 1267 1268 1269 |
allocated = 0; INIT_LIST_HEAD(&surplus_list); ret = -ENOMEM; retry: spin_unlock(&hugetlb_lock); for (i = 0; i < needed; i++) { |
bf50bab2b hugetlb: add allo... |
1270 |
page = alloc_buddy_huge_page(h, NUMA_NO_NODE); |
28073b02b mm: hugetlb: defe... |
1271 1272 1273 1274 |
if (!page) { alloc_ok = false; break; } |
e4e574b76 hugetlb: Try to g... |
1275 1276 |
list_add(&page->lru, &surplus_list); } |
28073b02b mm: hugetlb: defe... |
1277 |
allocated += i; |
e4e574b76 hugetlb: Try to g... |
1278 1279 1280 1281 1282 1283 |
/* * 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 ... |
1284 1285 |
needed = (h->resv_huge_pages + delta) - (h->free_huge_pages + allocated); |
28073b02b mm: hugetlb: defe... |
1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 |
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... |
1296 1297 |
/* * The surplus_list now contains _at_least_ the number of extra pages |
25985edce Fix common misspe... |
1298 |
* needed to accommodate the reservation. Add the appropriate number |
e4e574b76 hugetlb: Try to g... |
1299 |
* of pages to the hugetlb pool and free the extras back to the buddy |
ac09b3a15 hugetlb: close a ... |
1300 1301 1302 |
* 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... |
1303 1304 |
*/ needed += allocated; |
a55164389 hugetlb: modular ... |
1305 |
h->resv_huge_pages += delta; |
e4e574b76 hugetlb: Try to g... |
1306 |
ret = 0; |
a9869b837 hugetlb: move ref... |
1307 |
|
19fc3f0ac hugetlb: decrease... |
1308 |
/* Free the needed pages to the hugetlb pool */ |
e4e574b76 hugetlb: Try to g... |
1309 |
list_for_each_entry_safe(page, tmp, &surplus_list, lru) { |
19fc3f0ac hugetlb: decrease... |
1310 1311 |
if ((--needed) < 0) break; |
a9869b837 hugetlb: move ref... |
1312 1313 1314 1315 1316 |
/* * 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... |
1317 |
VM_BUG_ON_PAGE(page_count(page), page); |
a55164389 hugetlb: modular ... |
1318 |
enqueue_huge_page(h, page); |
19fc3f0ac hugetlb: decrease... |
1319 |
} |
28073b02b mm: hugetlb: defe... |
1320 |
free: |
b0365c8d0 mm: hugetlb: fix ... |
1321 |
spin_unlock(&hugetlb_lock); |
19fc3f0ac hugetlb: decrease... |
1322 1323 |
/* Free unnecessary surplus pages to the buddy allocator */ |
c0d934ba2 mm, hugetlb: remo... |
1324 1325 |
list_for_each_entry_safe(page, tmp, &surplus_list, lru) put_page(page); |
a9869b837 hugetlb: move ref... |
1326 |
spin_lock(&hugetlb_lock); |
e4e574b76 hugetlb: Try to g... |
1327 1328 1329 1330 1331 1332 1333 1334 |
return ret; } /* * When releasing a hugetlb pool reservation, any surplus pages that were * allocated to satisfy the reservation must be explicitly freed if they were * never used. |
685f34570 hugetlb: use free... |
1335 |
* Called with hugetlb_lock held. |
e4e574b76 hugetlb: Try to g... |
1336 |
*/ |
a55164389 hugetlb: modular ... |
1337 1338 |
static void return_unused_surplus_pages(struct hstate *h, unsigned long unused_resv_pages) |
e4e574b76 hugetlb: Try to g... |
1339 |
{ |
e4e574b76 hugetlb: Try to g... |
1340 |
unsigned long nr_pages; |
ac09b3a15 hugetlb: close a ... |
1341 |
/* Uncommit the reservation */ |
a55164389 hugetlb: modular ... |
1342 |
h->resv_huge_pages -= unused_resv_pages; |
ac09b3a15 hugetlb: close a ... |
1343 |
|
aa888a749 hugetlb: support ... |
1344 |
/* Cannot return gigantic pages currently */ |
bae7f4ae1 hugetlb: add hsta... |
1345 |
if (hstate_is_gigantic(h)) |
aa888a749 hugetlb: support ... |
1346 |
return; |
a55164389 hugetlb: modular ... |
1347 |
nr_pages = min(unused_resv_pages, h->surplus_huge_pages); |
e4e574b76 hugetlb: Try to g... |
1348 |
|
685f34570 hugetlb: use free... |
1349 1350 |
/* * We want to release as many surplus pages as possible, spread |
9b5e5d0fd hugetlb: use only... |
1351 1352 1353 1354 1355 |
* evenly across all nodes with memory. Iterate across these nodes * until we can no longer free unreserved surplus pages. This occurs * when the nodes with surplus pages have no free pages. * free_pool_huge_page() will balance the the freed pages across the * on-line nodes with memory and will handle the hstate accounting. |
685f34570 hugetlb: use free... |
1356 1357 |
*/ while (nr_pages--) { |
8cebfcd07 hugetlb: use N_ME... |
1358 |
if (!free_pool_huge_page(h, &node_states[N_MEMORY], 1)) |
685f34570 hugetlb: use free... |
1359 |
break; |
7848a4bf5 mm/hugetlb.c: add... |
1360 |
cond_resched_lock(&hugetlb_lock); |
e4e574b76 hugetlb: Try to g... |
1361 1362 |
} } |
c37f9fb11 hugetlb: allow hu... |
1363 1364 1365 |
/* * Determine if the huge page at addr within the vma has an associated * reservation. Where it does not we will need to logically increase |
90481622d hugepages: fix us... |
1366 1367 1368 1369 1370 1371 |
* reservation and actually increase subpool usage before an allocation * can occur. Where any new reservation would be required the * reservation change is prepared, but not committed. Once the page * has been allocated from the subpool and instantiated the change should * be committed via vma_commit_reservation. No action is required on * failure. |
c37f9fb11 hugetlb: allow hu... |
1372 |
*/ |
e2f17d945 hugetlb: chg cann... |
1373 |
static long vma_needs_reservation(struct hstate *h, |
a55164389 hugetlb: modular ... |
1374 |
struct vm_area_struct *vma, unsigned long addr) |
c37f9fb11 hugetlb: allow hu... |
1375 |
{ |
4e35f4838 mm, hugetlb: use ... |
1376 1377 1378 |
struct resv_map *resv; pgoff_t idx; long chg; |
c37f9fb11 hugetlb: allow hu... |
1379 |
|
4e35f4838 mm, hugetlb: use ... |
1380 1381 |
resv = vma_resv_map(vma); if (!resv) |
84afd99b8 hugetlb reservati... |
1382 |
return 1; |
c37f9fb11 hugetlb: allow hu... |
1383 |
|
4e35f4838 mm, hugetlb: use ... |
1384 1385 |
idx = vma_hugecache_offset(h, vma, addr); chg = region_chg(resv, idx, idx + 1); |
84afd99b8 hugetlb reservati... |
1386 |
|
4e35f4838 mm, hugetlb: use ... |
1387 1388 1389 1390 |
if (vma->vm_flags & VM_MAYSHARE) return chg; else return chg < 0 ? chg : 0; |
c37f9fb11 hugetlb: allow hu... |
1391 |
} |
a55164389 hugetlb: modular ... |
1392 1393 |
static void vma_commit_reservation(struct hstate *h, struct vm_area_struct *vma, unsigned long addr) |
c37f9fb11 hugetlb: allow hu... |
1394 |
{ |
4e35f4838 mm, hugetlb: use ... |
1395 1396 |
struct resv_map *resv; pgoff_t idx; |
84afd99b8 hugetlb reservati... |
1397 |
|
4e35f4838 mm, hugetlb: use ... |
1398 1399 1400 |
resv = vma_resv_map(vma); if (!resv) return; |
84afd99b8 hugetlb reservati... |
1401 |
|
4e35f4838 mm, hugetlb: use ... |
1402 1403 |
idx = vma_hugecache_offset(h, vma, addr); region_add(resv, idx, idx + 1); |
c37f9fb11 hugetlb: allow hu... |
1404 |
} |
a1e78772d hugetlb: reserve ... |
1405 |
static struct page *alloc_huge_page(struct vm_area_struct *vma, |
04f2cbe35 hugetlb: guarante... |
1406 |
unsigned long addr, int avoid_reserve) |
1da177e4c Linux-2.6.12-rc2 |
1407 |
{ |
90481622d hugepages: fix us... |
1408 |
struct hugepage_subpool *spool = subpool_vma(vma); |
a55164389 hugetlb: modular ... |
1409 |
struct hstate *h = hstate_vma(vma); |
348ea204c hugetlb: split al... |
1410 |
struct page *page; |
e2f17d945 hugetlb: chg cann... |
1411 |
long chg; |
6d76dcf40 hugetlb/cgroup: a... |
1412 1413 |
int ret, idx; struct hugetlb_cgroup *h_cg; |
a1e78772d hugetlb: reserve ... |
1414 |
|
6d76dcf40 hugetlb/cgroup: a... |
1415 |
idx = hstate_index(h); |
a1e78772d hugetlb: reserve ... |
1416 |
/* |
90481622d hugepages: fix us... |
1417 1418 1419 1420 1421 1422 |
* Processes that did not create the mapping will have no * reserves and will not have accounted against subpool * limit. Check that the subpool limit can be made before * satisfying the allocation MAP_NORESERVE mappings may also * need pages and subpool limit allocated allocated if no reserve * mapping overlaps. |
a1e78772d hugetlb: reserve ... |
1423 |
*/ |
a55164389 hugetlb: modular ... |
1424 |
chg = vma_needs_reservation(h, vma, addr); |
c37f9fb11 hugetlb: allow hu... |
1425 |
if (chg < 0) |
76dcee75c hugetlb: don't us... |
1426 |
return ERR_PTR(-ENOMEM); |
8bb3f12e7 mm, hugetlb: fix ... |
1427 |
if (chg || avoid_reserve) |
1c5ecae3a hugetlbfs: add mi... |
1428 |
if (hugepage_subpool_get_pages(spool, 1) < 0) |
76dcee75c hugetlb: don't us... |
1429 |
return ERR_PTR(-ENOSPC); |
1da177e4c Linux-2.6.12-rc2 |
1430 |
|
6d76dcf40 hugetlb/cgroup: a... |
1431 |
ret = hugetlb_cgroup_charge_cgroup(idx, pages_per_huge_page(h), &h_cg); |
8f34af6f9 mm, hugetlb: move... |
1432 1433 |
if (ret) goto out_subpool_put; |
1da177e4c Linux-2.6.12-rc2 |
1434 |
spin_lock(&hugetlb_lock); |
af0ed73e6 mm, hugetlb: decr... |
1435 |
page = dequeue_huge_page_vma(h, vma, addr, avoid_reserve, chg); |
81a6fcae3 mm, hugetlb: clea... |
1436 |
if (!page) { |
94ae8ba71 hugetlb/cgroup: a... |
1437 |
spin_unlock(&hugetlb_lock); |
bf50bab2b hugetlb: add allo... |
1438 |
page = alloc_buddy_huge_page(h, NUMA_NO_NODE); |
8f34af6f9 mm, hugetlb: move... |
1439 1440 |
if (!page) goto out_uncharge_cgroup; |
79dbb2368 hugetlb: move all... |
1441 1442 |
spin_lock(&hugetlb_lock); list_move(&page->lru, &h->hugepage_activelist); |
81a6fcae3 mm, hugetlb: clea... |
1443 |
/* Fall through */ |
68842c9b9 hugetlbfs: fix qu... |
1444 |
} |
81a6fcae3 mm, hugetlb: clea... |
1445 1446 |
hugetlb_cgroup_commit_charge(idx, pages_per_huge_page(h), h_cg, page); spin_unlock(&hugetlb_lock); |
348ea204c hugetlb: split al... |
1447 |
|
90481622d hugepages: fix us... |
1448 |
set_page_private(page, (unsigned long)spool); |
90d8b7e61 hugetlb: enforce ... |
1449 |
|
a55164389 hugetlb: modular ... |
1450 |
vma_commit_reservation(h, vma, addr); |
90d8b7e61 hugetlb: enforce ... |
1451 |
return page; |
8f34af6f9 mm, hugetlb: move... |
1452 1453 1454 1455 1456 1457 1458 |
out_uncharge_cgroup: hugetlb_cgroup_uncharge_cgroup(idx, pages_per_huge_page(h), h_cg); out_subpool_put: if (chg || avoid_reserve) hugepage_subpool_put_pages(spool, 1); return ERR_PTR(-ENOSPC); |
b45b5bd65 [PATCH] hugepage:... |
1459 |
} |
74060e4d7 mm: mbind: add hu... |
1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 |
/* * 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... |
1473 |
int __weak alloc_bootmem_huge_page(struct hstate *h) |
aa888a749 hugetlb: support ... |
1474 1475 |
{ struct huge_bootmem_page *m; |
b22610268 mm, hugetlb: fix ... |
1476 |
int nr_nodes, node; |
aa888a749 hugetlb: support ... |
1477 |
|
b22610268 mm, hugetlb: fix ... |
1478 |
for_each_node_mask_to_alloc(h, nr_nodes, node, &node_states[N_MEMORY]) { |
aa888a749 hugetlb: support ... |
1479 |
void *addr; |
8b89a1169 mm/hugetlb.c: use... |
1480 1481 1482 |
addr = memblock_virt_alloc_try_nid_nopanic( huge_page_size(h), huge_page_size(h), 0, BOOTMEM_ALLOC_ACCESSIBLE, node); |
aa888a749 hugetlb: support ... |
1483 1484 1485 1486 1487 1488 1489 |
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... |
1490 |
goto found; |
aa888a749 hugetlb: support ... |
1491 |
} |
aa888a749 hugetlb: support ... |
1492 1493 1494 1495 |
} return 0; found: |
df994ead5 hugetlb: alloc_bo... |
1496 |
BUG_ON(!IS_ALIGNED(virt_to_phys(m), huge_page_size(h))); |
aa888a749 hugetlb: support ... |
1497 1498 1499 1500 1501 |
/* Put them into a private list first because mem_map is not up yet */ list_add(&m->list, &huge_boot_pages); m->hstate = h; return 1; } |
f412c97ab mm, hugetlb: mark... |
1502 |
static void __init prep_compound_huge_page(struct page *page, int order) |
18229df5b hugetlb: pull gig... |
1503 1504 1505 1506 1507 1508 |
{ if (unlikely(order > (MAX_ORDER - 1))) prep_compound_gigantic_page(page, order); else prep_compound_page(page, order); } |
aa888a749 hugetlb: support ... |
1509 1510 1511 1512 1513 1514 |
/* 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 ... |
1515 |
struct hstate *h = m->hstate; |
ee8f248d2 hugetlb: add phys... |
1516 1517 1518 1519 |
struct page *page; #ifdef CONFIG_HIGHMEM page = pfn_to_page(m->phys >> PAGE_SHIFT); |
8b89a1169 mm/hugetlb.c: use... |
1520 1521 |
memblock_free_late(__pa(m), sizeof(struct huge_bootmem_page)); |
ee8f248d2 hugetlb: add phys... |
1522 1523 1524 |
#else page = virt_to_page(m); #endif |
aa888a749 hugetlb: support ... |
1525 |
WARN_ON(page_count(page) != 1); |
18229df5b hugetlb: pull gig... |
1526 |
prep_compound_huge_page(page, h->order); |
ef5a22be2 mm: hugetlb: init... |
1527 |
WARN_ON(PageReserved(page)); |
aa888a749 hugetlb: support ... |
1528 |
prep_new_huge_page(h, page, page_to_nid(page)); |
b0320c7b7 mm: fix negative ... |
1529 1530 1531 1532 1533 1534 |
/* * 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... |
1535 |
if (hstate_is_gigantic(h)) |
3dcc0571c mm: correctly upd... |
1536 |
adjust_managed_page_count(page, 1 << h->order); |
aa888a749 hugetlb: support ... |
1537 1538 |
} } |
8faa8b077 hugetlb: support ... |
1539 |
static void __init hugetlb_hstate_alloc_pages(struct hstate *h) |
1da177e4c Linux-2.6.12-rc2 |
1540 1541 |
{ unsigned long i; |
a55164389 hugetlb: modular ... |
1542 |
|
e5ff21594 hugetlb: multiple... |
1543 |
for (i = 0; i < h->max_huge_pages; ++i) { |
bae7f4ae1 hugetlb: add hsta... |
1544 |
if (hstate_is_gigantic(h)) { |
aa888a749 hugetlb: support ... |
1545 1546 |
if (!alloc_bootmem_huge_page(h)) break; |
9b5e5d0fd hugetlb: use only... |
1547 |
} else if (!alloc_fresh_huge_page(h, |
8cebfcd07 hugetlb: use N_ME... |
1548 |
&node_states[N_MEMORY])) |
1da177e4c Linux-2.6.12-rc2 |
1549 |
break; |
1da177e4c Linux-2.6.12-rc2 |
1550 |
} |
8faa8b077 hugetlb: support ... |
1551 |
h->max_huge_pages = i; |
e5ff21594 hugetlb: multiple... |
1552 1553 1554 1555 1556 1557 1558 |
} static void __init hugetlb_init_hstates(void) { struct hstate *h; for_each_hstate(h) { |
c791ad1e4 mm/hugetlb: intro... |
1559 1560 |
if (minimum_order > huge_page_order(h)) minimum_order = huge_page_order(h); |
8faa8b077 hugetlb: support ... |
1561 |
/* oversize hugepages were init'ed in early boot */ |
bae7f4ae1 hugetlb: add hsta... |
1562 |
if (!hstate_is_gigantic(h)) |
8faa8b077 hugetlb: support ... |
1563 |
hugetlb_hstate_alloc_pages(h); |
e5ff21594 hugetlb: multiple... |
1564 |
} |
c791ad1e4 mm/hugetlb: intro... |
1565 |
VM_BUG_ON(minimum_order == UINT_MAX); |
e5ff21594 hugetlb: multiple... |
1566 |
} |
4abd32dba hugetlb: printk c... |
1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 |
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... |
1577 1578 1579 1580 1581 |
static void __init report_hugepages(void) { struct hstate *h; for_each_hstate(h) { |
4abd32dba hugetlb: printk c... |
1582 |
char buf[32]; |
ffb22af5b mm/hugetlb.c: con... |
1583 1584 |
pr_info("HugeTLB registered %s page size, pre-allocated %ld pages ", |
4abd32dba hugetlb: printk c... |
1585 1586 |
memfmt(buf, huge_page_size(h)), h->free_huge_pages); |
e5ff21594 hugetlb: multiple... |
1587 1588 |
} } |
1da177e4c Linux-2.6.12-rc2 |
1589 |
#ifdef CONFIG_HIGHMEM |
6ae11b278 hugetlb: add node... |
1590 1591 |
static void try_to_free_low(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c Linux-2.6.12-rc2 |
1592 |
{ |
4415cc8df [PATCH] Hugepages... |
1593 |
int i; |
bae7f4ae1 hugetlb: add hsta... |
1594 |
if (hstate_is_gigantic(h)) |
aa888a749 hugetlb: support ... |
1595 |
return; |
6ae11b278 hugetlb: add node... |
1596 |
for_each_node_mask(i, *nodes_allowed) { |
1da177e4c Linux-2.6.12-rc2 |
1597 |
struct page *page, *next; |
a55164389 hugetlb: modular ... |
1598 1599 1600 |
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... |
1601 |
return; |
1da177e4c Linux-2.6.12-rc2 |
1602 1603 1604 |
if (PageHighMem(page)) continue; list_del(&page->lru); |
e5ff21594 hugetlb: multiple... |
1605 |
update_and_free_page(h, page); |
a55164389 hugetlb: modular ... |
1606 1607 |
h->free_huge_pages--; h->free_huge_pages_node[page_to_nid(page)]--; |
1da177e4c Linux-2.6.12-rc2 |
1608 1609 1610 1611 |
} } } #else |
6ae11b278 hugetlb: add node... |
1612 1613 |
static inline void try_to_free_low(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c Linux-2.6.12-rc2 |
1614 1615 1616 |
{ } #endif |
20a0307c0 mm: introduce Pag... |
1617 1618 1619 1620 1621 |
/* * 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... |
1622 1623 |
static int adjust_pool_surplus(struct hstate *h, nodemask_t *nodes_allowed, int delta) |
20a0307c0 mm: introduce Pag... |
1624 |
{ |
b22610268 mm, hugetlb: fix ... |
1625 |
int nr_nodes, node; |
20a0307c0 mm: introduce Pag... |
1626 1627 |
VM_BUG_ON(delta != -1 && delta != 1); |
20a0307c0 mm: introduce Pag... |
1628 |
|
b22610268 mm, hugetlb: fix ... |
1629 1630 1631 1632 |
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 ... |
1633 |
} |
b22610268 mm, hugetlb: fix ... |
1634 1635 1636 1637 1638 |
} 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 ... |
1639 |
} |
b22610268 mm, hugetlb: fix ... |
1640 1641 |
} return 0; |
20a0307c0 mm: introduce Pag... |
1642 |
|
b22610268 mm, hugetlb: fix ... |
1643 1644 1645 1646 |
found: h->surplus_huge_pages += delta; h->surplus_huge_pages_node[node] += delta; return 1; |
20a0307c0 mm: introduce Pag... |
1647 |
} |
a55164389 hugetlb: modular ... |
1648 |
#define persistent_huge_pages(h) (h->nr_huge_pages - h->surplus_huge_pages) |
6ae11b278 hugetlb: add node... |
1649 1650 |
static unsigned long set_max_huge_pages(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c Linux-2.6.12-rc2 |
1651 |
{ |
7893d1d50 hugetlb: Try to g... |
1652 |
unsigned long min_count, ret; |
1da177e4c Linux-2.6.12-rc2 |
1653 |
|
944d9fec8 hugetlb: add supp... |
1654 |
if (hstate_is_gigantic(h) && !gigantic_page_supported()) |
aa888a749 hugetlb: support ... |
1655 |
return h->max_huge_pages; |
7893d1d50 hugetlb: Try to g... |
1656 1657 1658 1659 |
/* * 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... |
1660 1661 1662 1663 1664 1665 |
* * We might race with alloc_buddy_huge_page() here and be unable * to convert a surplus huge page to a normal huge page. That is * not critical, though, it just means the overall size of the * pool might be one hugepage larger than it needs to be, but * within all the constraints specified by the sysctls. |
7893d1d50 hugetlb: Try to g... |
1666 |
*/ |
1da177e4c Linux-2.6.12-rc2 |
1667 |
spin_lock(&hugetlb_lock); |
a55164389 hugetlb: modular ... |
1668 |
while (h->surplus_huge_pages && count > persistent_huge_pages(h)) { |
6ae11b278 hugetlb: add node... |
1669 |
if (!adjust_pool_surplus(h, nodes_allowed, -1)) |
7893d1d50 hugetlb: Try to g... |
1670 1671 |
break; } |
a55164389 hugetlb: modular ... |
1672 |
while (count > persistent_huge_pages(h)) { |
7893d1d50 hugetlb: Try to g... |
1673 1674 1675 1676 1677 1678 |
/* * If this allocation races such that we no longer need the * page, free_huge_page will handle it by freeing the page * and reducing the surplus. */ spin_unlock(&hugetlb_lock); |
944d9fec8 hugetlb: add supp... |
1679 1680 1681 1682 |
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... |
1683 1684 1685 |
spin_lock(&hugetlb_lock); if (!ret) goto out; |
536240f2b hugetlb: abort a ... |
1686 1687 1688 |
/* Bail for signals. Probably ctrl-c from user */ if (signal_pending(current)) goto out; |
7893d1d50 hugetlb: Try to g... |
1689 |
} |
7893d1d50 hugetlb: Try to g... |
1690 1691 1692 1693 1694 1695 1696 |
/* * 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... |
1697 1698 1699 1700 1701 1702 1703 1704 |
* * By placing pages into the surplus state independent of the * overcommit value, we are allowing the surplus pool size to * exceed overcommit. There are few sane options here. Since * alloc_buddy_huge_page() is checking the global counter, * though, we'll note that we're not allowed to exceed surplus * and won't grow the pool anywhere else. Not until one of the * sysctls are changed, or the surplus pages go out of use. |
7893d1d50 hugetlb: Try to g... |
1705 |
*/ |
a55164389 hugetlb: modular ... |
1706 |
min_count = h->resv_huge_pages + h->nr_huge_pages - h->free_huge_pages; |
6b0c880df hugetlb: fix pool... |
1707 |
min_count = max(count, min_count); |
6ae11b278 hugetlb: add node... |
1708 |
try_to_free_low(h, min_count, nodes_allowed); |
a55164389 hugetlb: modular ... |
1709 |
while (min_count < persistent_huge_pages(h)) { |
6ae11b278 hugetlb: add node... |
1710 |
if (!free_pool_huge_page(h, nodes_allowed, 0)) |
1da177e4c Linux-2.6.12-rc2 |
1711 |
break; |
55f67141a mm: hugetlb: fix ... |
1712 |
cond_resched_lock(&hugetlb_lock); |
1da177e4c Linux-2.6.12-rc2 |
1713 |
} |
a55164389 hugetlb: modular ... |
1714 |
while (count < persistent_huge_pages(h)) { |
6ae11b278 hugetlb: add node... |
1715 |
if (!adjust_pool_surplus(h, nodes_allowed, 1)) |
7893d1d50 hugetlb: Try to g... |
1716 1717 1718 |
break; } out: |
a55164389 hugetlb: modular ... |
1719 |
ret = persistent_huge_pages(h); |
1da177e4c Linux-2.6.12-rc2 |
1720 |
spin_unlock(&hugetlb_lock); |
7893d1d50 hugetlb: Try to g... |
1721 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
1722 |
} |
a34378701 hugetlb: new sysf... |
1723 1724 1725 1726 1727 1728 1729 1730 1731 |
#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 ... |
1732 1733 1734 |
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... |
1735 1736 |
{ int i; |
9a3052306 hugetlb: add per ... |
1737 |
|
a34378701 hugetlb: new sysf... |
1738 |
for (i = 0; i < HUGE_MAX_HSTATE; i++) |
9a3052306 hugetlb: add per ... |
1739 1740 1741 |
if (hstate_kobjs[i] == kobj) { if (nidp) *nidp = NUMA_NO_NODE; |
a34378701 hugetlb: new sysf... |
1742 |
return &hstates[i]; |
9a3052306 hugetlb: add per ... |
1743 1744 1745 |
} return kobj_to_node_hstate(kobj, nidp); |
a34378701 hugetlb: new sysf... |
1746 |
} |
06808b082 hugetlb: derive h... |
1747 |
static ssize_t nr_hugepages_show_common(struct kobject *kobj, |
a34378701 hugetlb: new sysf... |
1748 1749 |
struct kobj_attribute *attr, char *buf) { |
9a3052306 hugetlb: add per ... |
1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 |
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... |
1762 |
} |
adbe8726d hugetlb: do not a... |
1763 |
|
238d3c13f mm, hugetlb: gene... |
1764 1765 1766 |
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... |
1767 1768 |
{ int err; |
bad44b5be mm: add gfp flags... |
1769 |
NODEMASK_ALLOC(nodemask_t, nodes_allowed, GFP_KERNEL | __GFP_NORETRY); |
a34378701 hugetlb: new sysf... |
1770 |
|
944d9fec8 hugetlb: add supp... |
1771 |
if (hstate_is_gigantic(h) && !gigantic_page_supported()) { |
adbe8726d hugetlb: do not a... |
1772 1773 1774 |
err = -EINVAL; goto out; } |
9a3052306 hugetlb: add per ... |
1775 1776 1777 1778 1779 1780 1781 |
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... |
1782 |
nodes_allowed = &node_states[N_MEMORY]; |
9a3052306 hugetlb: add per ... |
1783 1784 1785 1786 1787 1788 1789 1790 1791 |
} } 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... |
1792 |
nodes_allowed = &node_states[N_MEMORY]; |
9a3052306 hugetlb: add per ... |
1793 |
|
06808b082 hugetlb: derive h... |
1794 |
h->max_huge_pages = set_max_huge_pages(h, count, nodes_allowed); |
a34378701 hugetlb: new sysf... |
1795 |
|
8cebfcd07 hugetlb: use N_ME... |
1796 |
if (nodes_allowed != &node_states[N_MEMORY]) |
06808b082 hugetlb: derive h... |
1797 1798 1799 |
NODEMASK_FREE(nodes_allowed); return len; |
adbe8726d hugetlb: do not a... |
1800 1801 1802 |
out: NODEMASK_FREE(nodes_allowed); return err; |
06808b082 hugetlb: derive h... |
1803 |
} |
238d3c13f mm, hugetlb: gene... |
1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 |
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... |
1820 1821 1822 1823 1824 1825 1826 1827 1828 |
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... |
1829 |
return nr_hugepages_store_common(false, kobj, buf, len); |
a34378701 hugetlb: new sysf... |
1830 1831 |
} HSTATE_ATTR(nr_hugepages); |
06808b082 hugetlb: derive h... |
1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 |
#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... |
1847 |
return nr_hugepages_store_common(true, kobj, buf, len); |
06808b082 hugetlb: derive h... |
1848 1849 1850 |
} HSTATE_ATTR(nr_hugepages_mempolicy); #endif |
a34378701 hugetlb: new sysf... |
1851 1852 1853 |
static ssize_t nr_overcommit_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
9a3052306 hugetlb: add per ... |
1854 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701 hugetlb: new sysf... |
1855 1856 1857 |
return sprintf(buf, "%lu ", h->nr_overcommit_huge_pages); } |
adbe8726d hugetlb: do not a... |
1858 |
|
a34378701 hugetlb: new sysf... |
1859 1860 1861 1862 1863 |
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 ... |
1864 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701 hugetlb: new sysf... |
1865 |
|
bae7f4ae1 hugetlb: add hsta... |
1866 |
if (hstate_is_gigantic(h)) |
adbe8726d hugetlb: do not a... |
1867 |
return -EINVAL; |
3dbb95f78 mm: replace stric... |
1868 |
err = kstrtoul(buf, 10, &input); |
a34378701 hugetlb: new sysf... |
1869 |
if (err) |
73ae31e59 hugetlb: fix hand... |
1870 |
return err; |
a34378701 hugetlb: new sysf... |
1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 |
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 ... |
1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 |
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... |
1895 1896 1897 1898 1899 1900 |
} HSTATE_ATTR_RO(free_hugepages); static ssize_t resv_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
9a3052306 hugetlb: add per ... |
1901 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701 hugetlb: new sysf... |
1902 1903 1904 1905 1906 1907 1908 1909 |
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 ... |
1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 |
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... |
1922 1923 1924 1925 1926 1927 1928 1929 1930 |
} 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... |
1931 1932 1933 |
#ifdef CONFIG_NUMA &nr_hugepages_mempolicy_attr.attr, #endif |
a34378701 hugetlb: new sysf... |
1934 1935 1936 1937 1938 1939 |
NULL, }; static struct attribute_group hstate_attr_group = { .attrs = hstate_attrs, }; |
094e9539b hugetlb: fix sect... |
1940 1941 1942 |
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... |
1943 1944 |
{ int retval; |
972dc4de1 hugetlb: add an i... |
1945 |
int hi = hstate_index(h); |
a34378701 hugetlb: new sysf... |
1946 |
|
9a3052306 hugetlb: add per ... |
1947 1948 |
hstate_kobjs[hi] = kobject_create_and_add(h->name, parent); if (!hstate_kobjs[hi]) |
a34378701 hugetlb: new sysf... |
1949 |
return -ENOMEM; |
9a3052306 hugetlb: add per ... |
1950 |
retval = sysfs_create_group(hstate_kobjs[hi], hstate_attr_group); |
a34378701 hugetlb: new sysf... |
1951 |
if (retval) |
9a3052306 hugetlb: add per ... |
1952 |
kobject_put(hstate_kobjs[hi]); |
a34378701 hugetlb: new sysf... |
1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 |
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 ... |
1967 1968 |
err = hugetlb_sysfs_add_hstate(h, hugepages_kobj, hstate_kobjs, &hstate_attr_group); |
a34378701 hugetlb: new sysf... |
1969 |
if (err) |
ffb22af5b mm/hugetlb.c: con... |
1970 |
pr_err("Hugetlb: Unable to add hstate %s", h->name); |
a34378701 hugetlb: new sysf... |
1971 1972 |
} } |
9a3052306 hugetlb: add per ... |
1973 1974 1975 1976 |
#ifdef CONFIG_NUMA /* * node_hstate/s - associate per node hstate attributes, via their kobjects, |
10fbcf4c6 convert 'memory' ... |
1977 1978 1979 |
* 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 ... |
1980 1981 1982 1983 1984 1985 1986 1987 1988 |
* the base kernel, on the hugetlb module. */ struct node_hstate { struct kobject *hugepages_kobj; struct kobject *hstate_kobjs[HUGE_MAX_HSTATE]; }; struct node_hstate node_hstates[MAX_NUMNODES]; /* |
10fbcf4c6 convert 'memory' ... |
1989 |
* A subset of global hstate attributes for node devices |
9a3052306 hugetlb: add per ... |
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 |
*/ 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' ... |
2003 |
* kobj_to_node_hstate - lookup global hstate for node device hstate attr kobj. |
9a3052306 hugetlb: add per ... |
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 |
* 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' ... |
2026 |
* Unregister hstate attributes from a single node device. |
9a3052306 hugetlb: add per ... |
2027 2028 |
* No-op if no hstate attributes attached. */ |
3cd8b44fa hugetlb: fix spar... |
2029 |
static void hugetlb_unregister_node(struct node *node) |
9a3052306 hugetlb: add per ... |
2030 2031 |
{ struct hstate *h; |
10fbcf4c6 convert 'memory' ... |
2032 |
struct node_hstate *nhs = &node_hstates[node->dev.id]; |
9a3052306 hugetlb: add per ... |
2033 2034 |
if (!nhs->hugepages_kobj) |
9b5e5d0fd hugetlb: use only... |
2035 |
return; /* no hstate attributes */ |
9a3052306 hugetlb: add per ... |
2036 |
|
972dc4de1 hugetlb: add an i... |
2037 2038 2039 2040 2041 |
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 ... |
2042 |
} |
972dc4de1 hugetlb: add an i... |
2043 |
} |
9a3052306 hugetlb: add per ... |
2044 2045 2046 2047 2048 2049 |
kobject_put(nhs->hugepages_kobj); nhs->hugepages_kobj = NULL; } /* |
10fbcf4c6 convert 'memory' ... |
2050 |
* hugetlb module exit: unregister hstate attributes from node devices |
9a3052306 hugetlb: add per ... |
2051 2052 2053 2054 2055 2056 2057 |
* that have them. */ static void hugetlb_unregister_all_nodes(void) { int nid; /* |
10fbcf4c6 convert 'memory' ... |
2058 |
* disable node device registrations. |
9a3052306 hugetlb: add per ... |
2059 2060 2061 2062 2063 2064 2065 |
*/ register_hugetlbfs_with_node(NULL, NULL); /* * remove hstate attributes from any nodes that have them. */ for (nid = 0; nid < nr_node_ids; nid++) |
8732794b1 numa: convert sta... |
2066 |
hugetlb_unregister_node(node_devices[nid]); |
9a3052306 hugetlb: add per ... |
2067 2068 2069 |
} /* |
10fbcf4c6 convert 'memory' ... |
2070 |
* Register hstate attributes for a single node device. |
9a3052306 hugetlb: add per ... |
2071 2072 |
* No-op if attributes already registered. */ |
3cd8b44fa hugetlb: fix spar... |
2073 |
static void hugetlb_register_node(struct node *node) |
9a3052306 hugetlb: add per ... |
2074 2075 |
{ struct hstate *h; |
10fbcf4c6 convert 'memory' ... |
2076 |
struct node_hstate *nhs = &node_hstates[node->dev.id]; |
9a3052306 hugetlb: add per ... |
2077 2078 2079 2080 2081 2082 |
int err; if (nhs->hugepages_kobj) return; /* already allocated */ nhs->hugepages_kobj = kobject_create_and_add("hugepages", |
10fbcf4c6 convert 'memory' ... |
2083 |
&node->dev.kobj); |
9a3052306 hugetlb: add per ... |
2084 2085 2086 2087 2088 2089 2090 2091 |
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... |
2092 2093 2094 |
pr_err("Hugetlb: Unable to add hstate %s for node %d ", h->name, node->dev.id); |
9a3052306 hugetlb: add per ... |
2095 2096 2097 2098 2099 2100 2101 |
hugetlb_unregister_node(node); break; } } } /* |
9b5e5d0fd hugetlb: use only... |
2102 |
* hugetlb init time: register hstate attributes for all registered node |
10fbcf4c6 convert 'memory' ... |
2103 2104 |
* devices of nodes that have memory. All on-line nodes should have * registered their associated device by this time. |
9a3052306 hugetlb: add per ... |
2105 |
*/ |
7d9ca0004 hugetlb: hugetlb_... |
2106 |
static void __init hugetlb_register_all_nodes(void) |
9a3052306 hugetlb: add per ... |
2107 2108 |
{ int nid; |
8cebfcd07 hugetlb: use N_ME... |
2109 |
for_each_node_state(nid, N_MEMORY) { |
8732794b1 numa: convert sta... |
2110 |
struct node *node = node_devices[nid]; |
10fbcf4c6 convert 'memory' ... |
2111 |
if (node->dev.id == nid) |
9a3052306 hugetlb: add per ... |
2112 2113 2114 2115 |
hugetlb_register_node(node); } /* |
10fbcf4c6 convert 'memory' ... |
2116 |
* Let the node device driver know we're here so it can |
9a3052306 hugetlb: add per ... |
2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 |
* [un]register hstate attributes on node hotplug. */ register_hugetlbfs_with_node(hugetlb_register_node, hugetlb_unregister_node); } #else /* !CONFIG_NUMA */ static struct hstate *kobj_to_node_hstate(struct kobject *kobj, int *nidp) { BUG(); if (nidp) *nidp = -1; return NULL; } static void hugetlb_unregister_all_nodes(void) { } static void hugetlb_register_all_nodes(void) { } #endif |
a34378701 hugetlb: new sysf... |
2137 2138 2139 |
static void __exit hugetlb_exit(void) { struct hstate *h; |
9a3052306 hugetlb: add per ... |
2140 |
hugetlb_unregister_all_nodes(); |
a34378701 hugetlb: new sysf... |
2141 |
for_each_hstate(h) { |
972dc4de1 hugetlb: add an i... |
2142 |
kobject_put(hstate_kobjs[hstate_index(h)]); |
a34378701 hugetlb: new sysf... |
2143 2144 2145 |
} kobject_put(hugepages_kobj); |
8382d914e mm, hugetlb: impr... |
2146 |
kfree(htlb_fault_mutex_table); |
a34378701 hugetlb: new sysf... |
2147 2148 2149 2150 2151 |
} module_exit(hugetlb_exit); static int __init hugetlb_init(void) { |
8382d914e mm, hugetlb: impr... |
2152 |
int i; |
457c1b27e hugetlb: ensure h... |
2153 |
if (!hugepages_supported()) |
0ef89d25d mm/hugetlb: don't... |
2154 |
return 0; |
a34378701 hugetlb: new sysf... |
2155 |
|
e11bfbfcb hugetlb: override... |
2156 2157 2158 2159 |
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... |
2160 |
} |
972dc4de1 hugetlb: add an i... |
2161 |
default_hstate_idx = hstate_index(size_to_hstate(default_hstate_size)); |
e11bfbfcb hugetlb: override... |
2162 2163 |
if (default_hstate_max_huge_pages) default_hstate.max_huge_pages = default_hstate_max_huge_pages; |
a34378701 hugetlb: new sysf... |
2164 2165 |
hugetlb_init_hstates(); |
aa888a749 hugetlb: support ... |
2166 |
gather_bootmem_prealloc(); |
a34378701 hugetlb: new sysf... |
2167 2168 2169 |
report_hugepages(); hugetlb_sysfs_init(); |
9a3052306 hugetlb: add per ... |
2170 |
hugetlb_register_all_nodes(); |
7179e7bf4 mm/hugetlb: creat... |
2171 |
hugetlb_cgroup_file_init(); |
9a3052306 hugetlb: add per ... |
2172 |
|
8382d914e mm, hugetlb: impr... |
2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 |
#ifdef CONFIG_SMP num_fault_mutexes = roundup_pow_of_two(8 * num_possible_cpus()); #else num_fault_mutexes = 1; #endif htlb_fault_mutex_table = kmalloc(sizeof(struct mutex) * num_fault_mutexes, GFP_KERNEL); BUG_ON(!htlb_fault_mutex_table); for (i = 0; i < num_fault_mutexes; i++) mutex_init(&htlb_fault_mutex_table[i]); |
a34378701 hugetlb: new sysf... |
2184 2185 2186 2187 2188 2189 2190 2191 |
return 0; } module_init(hugetlb_init); /* Should be called on processing a hugepagesz=... option */ void __init hugetlb_add_hstate(unsigned order) { struct hstate *h; |
8faa8b077 hugetlb: support ... |
2192 |
unsigned long i; |
a34378701 hugetlb: new sysf... |
2193 |
if (size_to_hstate(PAGE_SIZE << order)) { |
ffb22af5b mm/hugetlb.c: con... |
2194 2195 |
pr_warning("hugepagesz= specified twice, ignoring "); |
a34378701 hugetlb: new sysf... |
2196 2197 |
return; } |
47d38344a hugetlb: rename m... |
2198 |
BUG_ON(hugetlb_max_hstate >= HUGE_MAX_HSTATE); |
a34378701 hugetlb: new sysf... |
2199 |
BUG_ON(order == 0); |
47d38344a hugetlb: rename m... |
2200 |
h = &hstates[hugetlb_max_hstate++]; |
a34378701 hugetlb: new sysf... |
2201 2202 |
h->order = order; h->mask = ~((1ULL << (order + PAGE_SHIFT)) - 1); |
8faa8b077 hugetlb: support ... |
2203 2204 2205 2206 |
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... |
2207 |
INIT_LIST_HEAD(&h->hugepage_activelist); |
8cebfcd07 hugetlb: use N_ME... |
2208 2209 |
h->next_nid_to_alloc = first_node(node_states[N_MEMORY]); h->next_nid_to_free = first_node(node_states[N_MEMORY]); |
a34378701 hugetlb: new sysf... |
2210 2211 |
snprintf(h->name, HSTATE_NAME_LEN, "hugepages-%lukB", huge_page_size(h)/1024); |
8faa8b077 hugetlb: support ... |
2212 |
|
a34378701 hugetlb: new sysf... |
2213 2214 |
parsed_hstate = h; } |
e11bfbfcb hugetlb: override... |
2215 |
static int __init hugetlb_nrpages_setup(char *s) |
a34378701 hugetlb: new sysf... |
2216 2217 |
{ unsigned long *mhp; |
8faa8b077 hugetlb: support ... |
2218 |
static unsigned long *last_mhp; |
a34378701 hugetlb: new sysf... |
2219 2220 |
/* |
47d38344a hugetlb: rename m... |
2221 |
* !hugetlb_max_hstate means we haven't parsed a hugepagesz= parameter yet, |
a34378701 hugetlb: new sysf... |
2222 2223 |
* so this hugepages= parameter goes to the "default hstate". */ |
47d38344a hugetlb: rename m... |
2224 |
if (!hugetlb_max_hstate) |
a34378701 hugetlb: new sysf... |
2225 2226 2227 |
mhp = &default_hstate_max_huge_pages; else mhp = &parsed_hstate->max_huge_pages; |
8faa8b077 hugetlb: support ... |
2228 |
if (mhp == last_mhp) { |
ffb22af5b mm/hugetlb.c: con... |
2229 2230 2231 |
pr_warning("hugepages= specified twice without " "interleaving hugepagesz=, ignoring "); |
8faa8b077 hugetlb: support ... |
2232 2233 |
return 1; } |
a34378701 hugetlb: new sysf... |
2234 2235 |
if (sscanf(s, "%lu", mhp) <= 0) *mhp = 0; |
8faa8b077 hugetlb: support ... |
2236 2237 2238 2239 2240 |
/* * 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... |
2241 |
if (hugetlb_max_hstate && parsed_hstate->order >= MAX_ORDER) |
8faa8b077 hugetlb: support ... |
2242 2243 2244 |
hugetlb_hstate_alloc_pages(parsed_hstate); last_mhp = mhp; |
a34378701 hugetlb: new sysf... |
2245 2246 |
return 1; } |
e11bfbfcb hugetlb: override... |
2247 2248 2249 2250 2251 2252 2253 2254 |
__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... |
2255 |
|
8a2134605 hugetlb: fix CONF... |
2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 |
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... |
2268 2269 2270 |
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 |
2271 |
{ |
e5ff21594 hugetlb: multiple... |
2272 |
struct hstate *h = &default_hstate; |
238d3c13f mm, hugetlb: gene... |
2273 |
unsigned long tmp = h->max_huge_pages; |
08d4a2465 hugetlb: check th... |
2274 |
int ret; |
e5ff21594 hugetlb: multiple... |
2275 |
|
457c1b27e hugetlb: ensure h... |
2276 2277 |
if (!hugepages_supported()) return -ENOTSUPP; |
e5ff21594 hugetlb: multiple... |
2278 2279 |
table->data = &tmp; table->maxlen = sizeof(unsigned long); |
08d4a2465 hugetlb: check th... |
2280 2281 2282 |
ret = proc_doulongvec_minmax(table, write, buffer, length, ppos); if (ret) goto out; |
e5ff21594 hugetlb: multiple... |
2283 |
|
238d3c13f mm, hugetlb: gene... |
2284 2285 2286 |
if (write) ret = __nr_hugepages_store_common(obey_mempolicy, h, NUMA_NO_NODE, tmp, *length); |
08d4a2465 hugetlb: check th... |
2287 2288 |
out: return ret; |
1da177e4c Linux-2.6.12-rc2 |
2289 |
} |
396faf030 Allow huge page a... |
2290 |
|
06808b082 hugetlb: derive h... |
2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 |
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... |
2307 |
int hugetlb_overcommit_handler(struct ctl_table *table, int write, |
8d65af789 sysctl: remove "s... |
2308 |
void __user *buffer, |
a3d0c6aa1 hugetlb: add lock... |
2309 2310 |
size_t *length, loff_t *ppos) { |
a55164389 hugetlb: modular ... |
2311 |
struct hstate *h = &default_hstate; |
e5ff21594 hugetlb: multiple... |
2312 |
unsigned long tmp; |
08d4a2465 hugetlb: check th... |
2313 |
int ret; |
e5ff21594 hugetlb: multiple... |
2314 |
|
457c1b27e hugetlb: ensure h... |
2315 2316 |
if (!hugepages_supported()) return -ENOTSUPP; |
c033a93c0 hugetlbfs: correc... |
2317 |
tmp = h->nr_overcommit_huge_pages; |
e5ff21594 hugetlb: multiple... |
2318 |
|
bae7f4ae1 hugetlb: add hsta... |
2319 |
if (write && hstate_is_gigantic(h)) |
adbe8726d hugetlb: do not a... |
2320 |
return -EINVAL; |
e5ff21594 hugetlb: multiple... |
2321 2322 |
table->data = &tmp; table->maxlen = sizeof(unsigned long); |
08d4a2465 hugetlb: check th... |
2323 2324 2325 |
ret = proc_doulongvec_minmax(table, write, buffer, length, ppos); if (ret) goto out; |
e5ff21594 hugetlb: multiple... |
2326 2327 2328 2329 2330 2331 |
if (write) { spin_lock(&hugetlb_lock); h->nr_overcommit_huge_pages = tmp; spin_unlock(&hugetlb_lock); } |
08d4a2465 hugetlb: check th... |
2332 2333 |
out: return ret; |
a3d0c6aa1 hugetlb: add lock... |
2334 |
} |
1da177e4c Linux-2.6.12-rc2 |
2335 |
#endif /* CONFIG_SYSCTL */ |
e1759c215 proc: switch /pro... |
2336 |
void hugetlb_report_meminfo(struct seq_file *m) |
1da177e4c Linux-2.6.12-rc2 |
2337 |
{ |
a55164389 hugetlb: modular ... |
2338 |
struct hstate *h = &default_hstate; |
457c1b27e hugetlb: ensure h... |
2339 2340 |
if (!hugepages_supported()) return; |
e1759c215 proc: switch /pro... |
2341 |
seq_printf(m, |
4f98a2fee vmscan: split LRU... |
2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 |
"HugePages_Total: %5lu " "HugePages_Free: %5lu " "HugePages_Rsvd: %5lu " "HugePages_Surp: %5lu " "Hugepagesize: %8lu kB ", |
a55164389 hugetlb: modular ... |
2352 2353 2354 2355 2356 |
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 |
2357 2358 2359 2360 |
} int hugetlb_report_node_meminfo(int nid, char *buf) { |
a55164389 hugetlb: modular ... |
2361 |
struct hstate *h = &default_hstate; |
457c1b27e hugetlb: ensure h... |
2362 2363 |
if (!hugepages_supported()) return 0; |
1da177e4c Linux-2.6.12-rc2 |
2364 2365 2366 |
return sprintf(buf, "Node %d HugePages_Total: %5u " |
a1de09195 hugetlb: indicate... |
2367 2368 2369 2370 |
"Node %d HugePages_Free: %5u " "Node %d HugePages_Surp: %5u ", |
a55164389 hugetlb: modular ... |
2371 2372 2373 |
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 |
2374 |
} |
949f7ec57 mm, hugetlb: incl... |
2375 2376 2377 2378 |
void hugetlb_show_meminfo(void) { struct hstate *h; int nid; |
457c1b27e hugetlb: ensure h... |
2379 2380 |
if (!hugepages_supported()) return; |
949f7ec57 mm, hugetlb: incl... |
2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 |
for_each_node_state(nid, N_MEMORY) for_each_hstate(h) pr_info("Node %d hugepages_total=%u hugepages_free=%u hugepages_surp=%u hugepages_size=%lukB ", nid, h->nr_huge_pages_node[nid], h->free_huge_pages_node[nid], h->surplus_huge_pages_node[nid], 1UL << (huge_page_order(h) + PAGE_SHIFT - 10)); } |
1da177e4c Linux-2.6.12-rc2 |
2391 2392 2393 |
/* Return the number pages of memory we physically have, in PAGE_SIZE units. */ unsigned long hugetlb_total_pages(void) { |
d00285884 mm/hugetlb: fix t... |
2394 2395 2396 2397 2398 2399 |
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 |
2400 |
} |
1da177e4c Linux-2.6.12-rc2 |
2401 |
|
a55164389 hugetlb: modular ... |
2402 |
static int hugetlb_acct_memory(struct hstate *h, long delta) |
fc1b8a73d hugetlb: move hug... |
2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 |
{ 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 ... |
2425 |
if (gather_surplus_pages(h, delta) < 0) |
fc1b8a73d hugetlb: move hug... |
2426 |
goto out; |
a55164389 hugetlb: modular ... |
2427 2428 |
if (delta > cpuset_mems_nr(h->free_huge_pages_node)) { return_unused_surplus_pages(h, delta); |
fc1b8a73d hugetlb: move hug... |
2429 2430 2431 2432 2433 2434 |
goto out; } } ret = 0; if (delta < 0) |
a55164389 hugetlb: modular ... |
2435 |
return_unused_surplus_pages(h, (unsigned long) -delta); |
fc1b8a73d hugetlb: move hug... |
2436 2437 2438 2439 2440 |
out: spin_unlock(&hugetlb_lock); return ret; } |
84afd99b8 hugetlb reservati... |
2441 2442 |
static void hugetlb_vm_op_open(struct vm_area_struct *vma) { |
f522c3ac0 mm, hugetlb: chan... |
2443 |
struct resv_map *resv = vma_resv_map(vma); |
84afd99b8 hugetlb reservati... |
2444 2445 2446 2447 2448 |
/* * 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... |
2449 |
* has a reference to the reservation map it cannot disappear until |
84afd99b8 hugetlb reservati... |
2450 2451 2452 |
* after this open call completes. It is therefore safe to take a * new reference here without additional locking. */ |
4e35f4838 mm, hugetlb: use ... |
2453 |
if (resv && is_vma_resv_set(vma, HPAGE_RESV_OWNER)) |
f522c3ac0 mm, hugetlb: chan... |
2454 |
kref_get(&resv->refs); |
84afd99b8 hugetlb reservati... |
2455 |
} |
a1e78772d hugetlb: reserve ... |
2456 2457 |
static void hugetlb_vm_op_close(struct vm_area_struct *vma) { |
a55164389 hugetlb: modular ... |
2458 |
struct hstate *h = hstate_vma(vma); |
f522c3ac0 mm, hugetlb: chan... |
2459 |
struct resv_map *resv = vma_resv_map(vma); |
90481622d hugepages: fix us... |
2460 |
struct hugepage_subpool *spool = subpool_vma(vma); |
4e35f4838 mm, hugetlb: use ... |
2461 |
unsigned long reserve, start, end; |
1c5ecae3a hugetlbfs: add mi... |
2462 |
long gbl_reserve; |
84afd99b8 hugetlb reservati... |
2463 |
|
4e35f4838 mm, hugetlb: use ... |
2464 2465 |
if (!resv || !is_vma_resv_set(vma, HPAGE_RESV_OWNER)) return; |
84afd99b8 hugetlb reservati... |
2466 |
|
4e35f4838 mm, hugetlb: use ... |
2467 2468 |
start = vma_hugecache_offset(h, vma, vma->vm_start); end = vma_hugecache_offset(h, vma, vma->vm_end); |
84afd99b8 hugetlb reservati... |
2469 |
|
4e35f4838 mm, hugetlb: use ... |
2470 |
reserve = (end - start) - region_count(resv, start, end); |
84afd99b8 hugetlb reservati... |
2471 |
|
4e35f4838 mm, hugetlb: use ... |
2472 2473 2474 |
kref_put(&resv->refs, resv_map_release); if (reserve) { |
1c5ecae3a hugetlbfs: add mi... |
2475 2476 2477 2478 2479 2480 |
/* * 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... |
2481 |
} |
a1e78772d hugetlb: reserve ... |
2482 |
} |
1da177e4c Linux-2.6.12-rc2 |
2483 2484 2485 2486 2487 2488 |
/* * 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... |
2489 |
static int hugetlb_vm_op_fault(struct vm_area_struct *vma, struct vm_fault *vmf) |
1da177e4c Linux-2.6.12-rc2 |
2490 2491 |
{ BUG(); |
d0217ac04 mm: fault feedbac... |
2492 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
2493 |
} |
f0f37e2f7 const: mark struc... |
2494 |
const struct vm_operations_struct hugetlb_vm_ops = { |
d0217ac04 mm: fault feedbac... |
2495 |
.fault = hugetlb_vm_op_fault, |
84afd99b8 hugetlb reservati... |
2496 |
.open = hugetlb_vm_op_open, |
a1e78772d hugetlb: reserve ... |
2497 |
.close = hugetlb_vm_op_close, |
1da177e4c Linux-2.6.12-rc2 |
2498 |
}; |
1e8f889b1 [PATCH] Hugetlb: ... |
2499 2500 |
static pte_t make_huge_pte(struct vm_area_struct *vma, struct page *page, int writable) |
63551ae0f [PATCH] Hugepage ... |
2501 2502 |
{ pte_t entry; |
1e8f889b1 [PATCH] Hugetlb: ... |
2503 |
if (writable) { |
106c992a5 mm/hugetlb: add m... |
2504 2505 |
entry = huge_pte_mkwrite(huge_pte_mkdirty(mk_huge_pte(page, vma->vm_page_prot))); |
63551ae0f [PATCH] Hugepage ... |
2506 |
} else { |
106c992a5 mm/hugetlb: add m... |
2507 2508 |
entry = huge_pte_wrprotect(mk_huge_pte(page, vma->vm_page_prot)); |
63551ae0f [PATCH] Hugepage ... |
2509 2510 2511 |
} entry = pte_mkyoung(entry); entry = pte_mkhuge(entry); |
d9ed9faac mm: add new arch_... |
2512 |
entry = arch_make_huge_pte(entry, vma, page, writable); |
63551ae0f [PATCH] Hugepage ... |
2513 2514 2515 |
return entry; } |
1e8f889b1 [PATCH] Hugetlb: ... |
2516 2517 2518 2519 |
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... |
2520 |
entry = huge_pte_mkwrite(huge_pte_mkdirty(huge_ptep_get(ptep))); |
32f84528f mm: hugetlb: fix ... |
2521 |
if (huge_ptep_set_access_flags(vma, address, ptep, entry, 1)) |
4b3073e1c MM: Pass a PTE po... |
2522 |
update_mmu_cache(vma, address, ptep); |
1e8f889b1 [PATCH] Hugetlb: ... |
2523 |
} |
4a705fef9 hugetlb: fix copy... |
2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 |
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: ... |
2549 |
|
63551ae0f [PATCH] Hugepage ... |
2550 2551 2552 2553 2554 |
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... |
2555 |
unsigned long addr; |
1e8f889b1 [PATCH] Hugetlb: ... |
2556 |
int cow; |
a55164389 hugetlb: modular ... |
2557 2558 |
struct hstate *h = hstate_vma(vma); unsigned long sz = huge_page_size(h); |
e8569dd29 mm/hugetlb.c: cal... |
2559 2560 2561 |
unsigned long mmun_start; /* For mmu_notifiers */ unsigned long mmun_end; /* For mmu_notifiers */ int ret = 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
2562 2563 |
cow = (vma->vm_flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE; |
63551ae0f [PATCH] Hugepage ... |
2564 |
|
e8569dd29 mm/hugetlb.c: cal... |
2565 2566 2567 2568 |
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 ... |
2569 |
for (addr = vma->vm_start; addr < vma->vm_end; addr += sz) { |
cb900f412 mm, hugetlb: conv... |
2570 |
spinlock_t *src_ptl, *dst_ptl; |
c74df32c7 [PATCH] mm: ptd_a... |
2571 2572 2573 |
src_pte = huge_pte_offset(src, addr); if (!src_pte) continue; |
a55164389 hugetlb: modular ... |
2574 |
dst_pte = huge_pte_alloc(dst, addr, sz); |
e8569dd29 mm/hugetlb.c: cal... |
2575 2576 2577 2578 |
if (!dst_pte) { ret = -ENOMEM; break; } |
c5c99429f fix hugepages lea... |
2579 2580 2581 2582 |
/* If the pagetables are shared don't copy or take references */ if (dst_pte == src_pte) continue; |
cb900f412 mm, hugetlb: conv... |
2583 2584 2585 |
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... |
2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 |
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... |
2604 |
if (cow) { |
7f2e9525b hugetlbfs: common... |
2605 |
huge_ptep_set_wrprotect(src, addr, src_pte); |
34ee645e8 mmu_notifier: cal... |
2606 2607 2608 |
mmu_notifier_invalidate_range(src, mmun_start, mmun_end); } |
0253d634e mm: hugetlb: fix ... |
2609 |
entry = huge_ptep_get(src_pte); |
1c59827d1 [PATCH] mm: huget... |
2610 2611 |
ptepage = pte_page(entry); get_page(ptepage); |
0fe6e20b9 hugetlb, rmap: ad... |
2612 |
page_dup_rmap(ptepage); |
1c59827d1 [PATCH] mm: huget... |
2613 2614 |
set_huge_pte_at(dst, addr, dst_pte, entry); } |
cb900f412 mm, hugetlb: conv... |
2615 2616 |
spin_unlock(src_ptl); spin_unlock(dst_ptl); |
63551ae0f [PATCH] Hugepage ... |
2617 |
} |
63551ae0f [PATCH] Hugepage ... |
2618 |
|
e8569dd29 mm/hugetlb.c: cal... |
2619 2620 2621 2622 |
if (cow) mmu_notifier_invalidate_range_end(src, mmun_start, mmun_end); return ret; |
63551ae0f [PATCH] Hugepage ... |
2623 |
} |
24669e584 hugetlb: use mmu_... |
2624 2625 2626 |
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 ... |
2627 |
{ |
24669e584 hugetlb: use mmu_... |
2628 |
int force_flush = 0; |
63551ae0f [PATCH] Hugepage ... |
2629 2630 |
struct mm_struct *mm = vma->vm_mm; unsigned long address; |
c7546f8f0 [PATCH] Fix hugep... |
2631 |
pte_t *ptep; |
63551ae0f [PATCH] Hugepage ... |
2632 |
pte_t pte; |
cb900f412 mm, hugetlb: conv... |
2633 |
spinlock_t *ptl; |
63551ae0f [PATCH] Hugepage ... |
2634 |
struct page *page; |
a55164389 hugetlb: modular ... |
2635 2636 |
struct hstate *h = hstate_vma(vma); unsigned long sz = huge_page_size(h); |
2ec74c3ef mm: move all mmu ... |
2637 2638 |
const unsigned long mmun_start = start; /* For mmu_notifiers */ const unsigned long mmun_end = end; /* For mmu_notifiers */ |
a55164389 hugetlb: modular ... |
2639 |
|
63551ae0f [PATCH] Hugepage ... |
2640 |
WARN_ON(!is_vm_hugetlb_page(vma)); |
a55164389 hugetlb: modular ... |
2641 2642 |
BUG_ON(start & ~huge_page_mask(h)); BUG_ON(end & ~huge_page_mask(h)); |
63551ae0f [PATCH] Hugepage ... |
2643 |
|
24669e584 hugetlb: use mmu_... |
2644 |
tlb_start_vma(tlb, vma); |
2ec74c3ef mm: move all mmu ... |
2645 |
mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); |
569f48b85 mm: hugetlb: fix ... |
2646 |
address = start; |
24669e584 hugetlb: use mmu_... |
2647 |
again: |
569f48b85 mm: hugetlb: fix ... |
2648 |
for (; address < end; address += sz) { |
c7546f8f0 [PATCH] Fix hugep... |
2649 |
ptep = huge_pte_offset(mm, address); |
4c8872659 [PATCH] hugetlb: ... |
2650 |
if (!ptep) |
c7546f8f0 [PATCH] Fix hugep... |
2651 |
continue; |
cb900f412 mm, hugetlb: conv... |
2652 |
ptl = huge_pte_lock(h, mm, ptep); |
39dde65c9 [PATCH] shared pa... |
2653 |
if (huge_pmd_unshare(mm, &address, ptep)) |
cb900f412 mm, hugetlb: conv... |
2654 |
goto unlock; |
39dde65c9 [PATCH] shared pa... |
2655 |
|
6629326b8 mm: hugetlb: clea... |
2656 2657 |
pte = huge_ptep_get(ptep); if (huge_pte_none(pte)) |
cb900f412 mm, hugetlb: conv... |
2658 |
goto unlock; |
6629326b8 mm: hugetlb: clea... |
2659 2660 |
/* |
9fbc1f635 mm/hugetlb: add m... |
2661 2662 |
* Migrating hugepage or HWPoisoned hugepage is already * unmapped and its refcount is dropped, so just clear pte here. |
6629326b8 mm: hugetlb: clea... |
2663 |
*/ |
9fbc1f635 mm/hugetlb: add m... |
2664 |
if (unlikely(!pte_present(pte))) { |
106c992a5 mm/hugetlb: add m... |
2665 |
huge_pte_clear(mm, address, ptep); |
cb900f412 mm, hugetlb: conv... |
2666 |
goto unlock; |
8c4894c6b hwpoison, hugetlb... |
2667 |
} |
6629326b8 mm: hugetlb: clea... |
2668 2669 |
page = pte_page(pte); |
04f2cbe35 hugetlb: guarante... |
2670 2671 2672 2673 2674 2675 |
/* * If a reference page is supplied, it is because a specific * page is being unmapped, not a range. Ensure the page we * are about to unmap is the actual page of interest. */ if (ref_page) { |
04f2cbe35 hugetlb: guarante... |
2676 |
if (page != ref_page) |
cb900f412 mm, hugetlb: conv... |
2677 |
goto unlock; |
04f2cbe35 hugetlb: guarante... |
2678 2679 2680 2681 2682 2683 2684 2685 |
/* * 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... |
2686 |
pte = huge_ptep_get_and_clear(mm, address, ptep); |
24669e584 hugetlb: use mmu_... |
2687 |
tlb_remove_tlb_entry(tlb, ptep, address); |
106c992a5 mm/hugetlb: add m... |
2688 |
if (huge_pte_dirty(pte)) |
6649a3863 [PATCH] hugetlb: ... |
2689 |
set_page_dirty(page); |
9e81130b7 mm: hugetlb: bail... |
2690 |
|
24669e584 hugetlb: use mmu_... |
2691 2692 |
page_remove_rmap(page); force_flush = !__tlb_remove_page(tlb, page); |
cb900f412 mm, hugetlb: conv... |
2693 |
if (force_flush) { |
569f48b85 mm: hugetlb: fix ... |
2694 |
address += sz; |
cb900f412 mm, hugetlb: conv... |
2695 |
spin_unlock(ptl); |
24669e584 hugetlb: use mmu_... |
2696 |
break; |
cb900f412 mm, hugetlb: conv... |
2697 |
} |
9e81130b7 mm: hugetlb: bail... |
2698 |
/* Bail out after unmapping reference page if supplied */ |
cb900f412 mm, hugetlb: conv... |
2699 2700 |
if (ref_page) { spin_unlock(ptl); |
9e81130b7 mm: hugetlb: bail... |
2701 |
break; |
cb900f412 mm, hugetlb: conv... |
2702 2703 2704 |
} unlock: spin_unlock(ptl); |
63551ae0f [PATCH] Hugepage ... |
2705 |
} |
24669e584 hugetlb: use mmu_... |
2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 |
/* * mmu_gather ran out of room to batch pages, we break out of * the PTE lock to avoid doing the potential expensive TLB invalidate * and page-free while holding it. */ if (force_flush) { force_flush = 0; tlb_flush_mmu(tlb); if (address < end && !ref_page) goto again; |
fe1668ae5 [PATCH] enforce p... |
2716 |
} |
2ec74c3ef mm: move all mmu ... |
2717 |
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
24669e584 hugetlb: use mmu_... |
2718 |
tlb_end_vma(tlb, vma); |
1da177e4c Linux-2.6.12-rc2 |
2719 |
} |
63551ae0f [PATCH] Hugepage ... |
2720 |
|
d833352a4 mm: hugetlbfs: cl... |
2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 |
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... |
2733 |
* is to clear it before releasing the i_mmap_rwsem. This works |
d833352a4 mm: hugetlbfs: cl... |
2734 |
* because in the context this is called, the VMA is about to be |
c8c06efa8 mm: convert i_mma... |
2735 |
* destroyed and the i_mmap_rwsem is held. |
d833352a4 mm: hugetlbfs: cl... |
2736 2737 2738 |
*/ vma->vm_flags &= ~VM_MAYSHARE; } |
502717f4e [PATCH] hugetlb: ... |
2739 |
void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start, |
04f2cbe35 hugetlb: guarante... |
2740 |
unsigned long end, struct page *ref_page) |
502717f4e [PATCH] hugetlb: ... |
2741 |
{ |
24669e584 hugetlb: use mmu_... |
2742 2743 2744 2745 |
struct mm_struct *mm; struct mmu_gather tlb; mm = vma->vm_mm; |
2b047252d Fix TLB gather vi... |
2746 |
tlb_gather_mmu(&tlb, mm, start, end); |
24669e584 hugetlb: use mmu_... |
2747 2748 |
__unmap_hugepage_range(&tlb, vma, start, end, ref_page); tlb_finish_mmu(&tlb, start, end); |
502717f4e [PATCH] hugetlb: ... |
2749 |
} |
04f2cbe35 hugetlb: guarante... |
2750 2751 2752 2753 2754 2755 |
/* * 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 ... |
2756 2757 |
static void unmap_ref_private(struct mm_struct *mm, struct vm_area_struct *vma, struct page *page, unsigned long address) |
04f2cbe35 hugetlb: guarante... |
2758 |
{ |
7526674de hugetlb: make unm... |
2759 |
struct hstate *h = hstate_vma(vma); |
04f2cbe35 hugetlb: guarante... |
2760 2761 |
struct vm_area_struct *iter_vma; struct address_space *mapping; |
04f2cbe35 hugetlb: guarante... |
2762 2763 2764 2765 2766 2767 |
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... |
2768 |
address = address & huge_page_mask(h); |
36e4f20af hugetlb: do not u... |
2769 2770 |
pgoff = ((address - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff; |
496ad9aa8 new helper: file_... |
2771 |
mapping = file_inode(vma->vm_file)->i_mapping; |
04f2cbe35 hugetlb: guarante... |
2772 |
|
4eb2b1dcd hugetlb: acquire ... |
2773 2774 2775 2776 2777 |
/* * 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... |
2778 |
i_mmap_lock_write(mapping); |
6b2dbba8b mm: replace vma p... |
2779 |
vma_interval_tree_foreach(iter_vma, &mapping->i_mmap, pgoff, pgoff) { |
04f2cbe35 hugetlb: guarante... |
2780 2781 2782 2783 2784 |
/* Do not unmap the current VMA */ if (iter_vma == vma) continue; /* |
51d3a065d mm: hugetlbfs: sk... |
2785 2786 2787 2788 2789 2790 2791 2792 |
* 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... |
2793 2794 2795 2796 2797 2798 2799 |
* 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_... |
2800 2801 |
unmap_hugepage_range(iter_vma, address, address + huge_page_size(h), page); |
04f2cbe35 hugetlb: guarante... |
2802 |
} |
83cde9e8b mm: use new helpe... |
2803 |
i_mmap_unlock_write(mapping); |
04f2cbe35 hugetlb: guarante... |
2804 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
2805 2806 |
/* * Hugetlb_cow() should be called with page lock of the original hugepage held. |
ef009b25f hugetlb: clarify ... |
2807 2808 2809 |
* 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... |
2810 |
*/ |
1e8f889b1 [PATCH] Hugetlb: ... |
2811 |
static int hugetlb_cow(struct mm_struct *mm, struct vm_area_struct *vma, |
04f2cbe35 hugetlb: guarante... |
2812 |
unsigned long address, pte_t *ptep, pte_t pte, |
cb900f412 mm, hugetlb: conv... |
2813 |
struct page *pagecache_page, spinlock_t *ptl) |
1e8f889b1 [PATCH] Hugetlb: ... |
2814 |
{ |
a55164389 hugetlb: modular ... |
2815 |
struct hstate *h = hstate_vma(vma); |
1e8f889b1 [PATCH] Hugetlb: ... |
2816 |
struct page *old_page, *new_page; |
ad4404a22 mm,hugetlb: simpl... |
2817 |
int ret = 0, outside_reserve = 0; |
2ec74c3ef mm: move all mmu ... |
2818 2819 |
unsigned long mmun_start; /* For mmu_notifiers */ unsigned long mmun_end; /* For mmu_notifiers */ |
1e8f889b1 [PATCH] Hugetlb: ... |
2820 2821 |
old_page = pte_page(pte); |
04f2cbe35 hugetlb: guarante... |
2822 |
retry_avoidcopy: |
1e8f889b1 [PATCH] Hugetlb: ... |
2823 2824 |
/* If no-one else is actually using this page, avoid the copy * and just make the page writable */ |
37a2140dc mm, hugetlb: do n... |
2825 2826 |
if (page_mapcount(old_page) == 1 && PageAnon(old_page)) { page_move_anon_rmap(old_page, vma, address); |
1e8f889b1 [PATCH] Hugetlb: ... |
2827 |
set_huge_ptep_writable(vma, address, ptep); |
83c54070e mm: fault feedbac... |
2828 |
return 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
2829 |
} |
04f2cbe35 hugetlb: guarante... |
2830 2831 2832 2833 2834 2835 2836 2837 2838 |
/* * 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... |
2839 |
if (is_vma_resv_set(vma, HPAGE_RESV_OWNER) && |
04f2cbe35 hugetlb: guarante... |
2840 2841 |
old_page != pagecache_page) outside_reserve = 1; |
1e8f889b1 [PATCH] Hugetlb: ... |
2842 |
page_cache_get(old_page); |
b76c8cfbf hugetlb: prevent ... |
2843 |
|
ad4404a22 mm,hugetlb: simpl... |
2844 2845 2846 2847 |
/* * 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... |
2848 |
spin_unlock(ptl); |
04f2cbe35 hugetlb: guarante... |
2849 |
new_page = alloc_huge_page(vma, address, outside_reserve); |
1e8f889b1 [PATCH] Hugetlb: ... |
2850 |
|
2fc39cec6 hugetlb: debit qu... |
2851 |
if (IS_ERR(new_page)) { |
04f2cbe35 hugetlb: guarante... |
2852 2853 2854 2855 2856 2857 2858 2859 |
/* * If a process owning a MAP_PRIVATE mapping fails to COW, * it is due to references held by a child and an insufficient * huge page pool. To guarantee the original mappers * reliability, unmap the page from child processes. The child * may get SIGKILLed if it later faults. */ if (outside_reserve) { |
ad4404a22 mm,hugetlb: simpl... |
2860 |
page_cache_release(old_page); |
04f2cbe35 hugetlb: guarante... |
2861 |
BUG_ON(huge_pte_none(pte)); |
2f4612af4 mm,hugetlb: make ... |
2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 |
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... |
2874 |
} |
ad4404a22 mm,hugetlb: simpl... |
2875 2876 2877 |
ret = (PTR_ERR(new_page) == -ENOMEM) ? VM_FAULT_OOM : VM_FAULT_SIGBUS; goto out_release_old; |
1e8f889b1 [PATCH] Hugetlb: ... |
2878 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
2879 2880 2881 2882 |
/* * When the original hugepage is shared one, it does not have * anon_vma prepared. */ |
44e2aa937 mm/hugetlb.c: add... |
2883 |
if (unlikely(anon_vma_prepare(vma))) { |
ad4404a22 mm,hugetlb: simpl... |
2884 2885 |
ret = VM_FAULT_OOM; goto out_release_all; |
44e2aa937 mm/hugetlb.c: add... |
2886 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
2887 |
|
47ad8475c thp: clear_copy_h... |
2888 2889 |
copy_user_huge_page(new_page, old_page, address, vma, pages_per_huge_page(h)); |
0ed361dec mm: fix PageUptod... |
2890 |
__SetPageUptodate(new_page); |
bcc542223 mm: hugetlb: intr... |
2891 |
set_page_huge_active(new_page); |
1e8f889b1 [PATCH] Hugetlb: ... |
2892 |
|
2ec74c3ef mm: move all mmu ... |
2893 2894 2895 |
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... |
2896 |
|
b76c8cfbf hugetlb: prevent ... |
2897 |
/* |
cb900f412 mm, hugetlb: conv... |
2898 |
* Retake the page table lock to check for racing updates |
b76c8cfbf hugetlb: prevent ... |
2899 2900 |
* before the page tables are altered */ |
cb900f412 mm, hugetlb: conv... |
2901 |
spin_lock(ptl); |
a55164389 hugetlb: modular ... |
2902 |
ptep = huge_pte_offset(mm, address & huge_page_mask(h)); |
a9af0c5df mm/hugetlb.c: add... |
2903 |
if (likely(ptep && pte_same(huge_ptep_get(ptep), pte))) { |
07443a85a mm, hugetlb: retu... |
2904 |
ClearPagePrivate(new_page); |
1e8f889b1 [PATCH] Hugetlb: ... |
2905 |
/* Break COW */ |
8fe627ec5 hugetlbfs: add mi... |
2906 |
huge_ptep_clear_flush(vma, address, ptep); |
34ee645e8 mmu_notifier: cal... |
2907 |
mmu_notifier_invalidate_range(mm, mmun_start, mmun_end); |
1e8f889b1 [PATCH] Hugetlb: ... |
2908 2909 |
set_huge_pte_at(mm, address, ptep, make_huge_pte(vma, new_page, 1)); |
0fe6e20b9 hugetlb, rmap: ad... |
2910 |
page_remove_rmap(old_page); |
cd67f0d2a hugetlb, rmap: us... |
2911 |
hugepage_add_new_anon_rmap(new_page, vma, address); |
1e8f889b1 [PATCH] Hugetlb: ... |
2912 2913 2914 |
/* Make the old page be freed below */ new_page = old_page; } |
cb900f412 mm, hugetlb: conv... |
2915 |
spin_unlock(ptl); |
2ec74c3ef mm: move all mmu ... |
2916 |
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
ad4404a22 mm,hugetlb: simpl... |
2917 |
out_release_all: |
1e8f889b1 [PATCH] Hugetlb: ... |
2918 |
page_cache_release(new_page); |
ad4404a22 mm,hugetlb: simpl... |
2919 |
out_release_old: |
1e8f889b1 [PATCH] Hugetlb: ... |
2920 |
page_cache_release(old_page); |
8312034f3 mm, hugetlb: grab... |
2921 |
|
ad4404a22 mm,hugetlb: simpl... |
2922 2923 |
spin_lock(ptl); /* Caller expects lock to be held */ return ret; |
1e8f889b1 [PATCH] Hugetlb: ... |
2924 |
} |
04f2cbe35 hugetlb: guarante... |
2925 |
/* Return the pagecache page at a given address within a VMA */ |
a55164389 hugetlb: modular ... |
2926 2927 |
static struct page *hugetlbfs_pagecache_page(struct hstate *h, struct vm_area_struct *vma, unsigned long address) |
04f2cbe35 hugetlb: guarante... |
2928 2929 |
{ struct address_space *mapping; |
e7c4b0bfd huge page private... |
2930 |
pgoff_t idx; |
04f2cbe35 hugetlb: guarante... |
2931 2932 |
mapping = vma->vm_file->f_mapping; |
a55164389 hugetlb: modular ... |
2933 |
idx = vma_hugecache_offset(h, vma, address); |
04f2cbe35 hugetlb: guarante... |
2934 2935 2936 |
return find_lock_page(mapping, idx); } |
3ae77f43b mm: hugetlbfs_pag... |
2937 2938 2939 2940 2941 |
/* * 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... |
2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 |
struct vm_area_struct *vma, unsigned long address) { struct address_space *mapping; pgoff_t idx; struct page *page; mapping = vma->vm_file->f_mapping; idx = vma_hugecache_offset(h, vma, address); page = find_get_page(mapping, idx); if (page) put_page(page); return page != NULL; } |
a1ed3dda0 MM: Make needless... |
2956 |
static int hugetlb_no_page(struct mm_struct *mm, struct vm_area_struct *vma, |
8382d914e mm, hugetlb: impr... |
2957 2958 |
struct address_space *mapping, pgoff_t idx, unsigned long address, pte_t *ptep, unsigned int flags) |
ac9b9c667 [PATCH] Fix handl... |
2959 |
{ |
a55164389 hugetlb: modular ... |
2960 |
struct hstate *h = hstate_vma(vma); |
ac9b9c667 [PATCH] Fix handl... |
2961 |
int ret = VM_FAULT_SIGBUS; |
409eb8c26 mm/hugetlb.c: und... |
2962 |
int anon_rmap = 0; |
4c8872659 [PATCH] hugetlb: ... |
2963 |
unsigned long size; |
4c8872659 [PATCH] hugetlb: ... |
2964 |
struct page *page; |
1e8f889b1 [PATCH] Hugetlb: ... |
2965 |
pte_t new_pte; |
cb900f412 mm, hugetlb: conv... |
2966 |
spinlock_t *ptl; |
4c8872659 [PATCH] hugetlb: ... |
2967 |
|
04f2cbe35 hugetlb: guarante... |
2968 2969 2970 |
/* * 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... |
2971 |
* COW. Warn that such a situation has occurred as it may not be obvious |
04f2cbe35 hugetlb: guarante... |
2972 2973 |
*/ if (is_vma_resv_set(vma, HPAGE_RESV_UNMAPPED)) { |
ffb22af5b mm/hugetlb.c: con... |
2974 2975 2976 |
pr_warning("PID %d killed due to inadequate hugepage pool ", current->pid); |
04f2cbe35 hugetlb: guarante... |
2977 2978 |
return ret; } |
4c8872659 [PATCH] hugetlb: ... |
2979 2980 2981 2982 |
/* * Use page lock to guard against racing truncation * before we get page_table_lock. */ |
6bda666a0 [PATCH] hugepages... |
2983 2984 2985 |
retry: page = find_lock_page(mapping, idx); if (!page) { |
a55164389 hugetlb: modular ... |
2986 |
size = i_size_read(mapping->host) >> huge_page_shift(h); |
ebed4bfc8 [PATCH] hugetlb: ... |
2987 2988 |
if (idx >= size) goto out; |
04f2cbe35 hugetlb: guarante... |
2989 |
page = alloc_huge_page(vma, address, 0); |
2fc39cec6 hugetlb: debit qu... |
2990 |
if (IS_ERR(page)) { |
76dcee75c hugetlb: don't us... |
2991 2992 2993 2994 2995 |
ret = PTR_ERR(page); if (ret == -ENOMEM) ret = VM_FAULT_OOM; else ret = VM_FAULT_SIGBUS; |
6bda666a0 [PATCH] hugepages... |
2996 2997 |
goto out; } |
47ad8475c thp: clear_copy_h... |
2998 |
clear_huge_page(page, address, pages_per_huge_page(h)); |
0ed361dec mm: fix PageUptod... |
2999 |
__SetPageUptodate(page); |
bcc542223 mm: hugetlb: intr... |
3000 |
set_page_huge_active(page); |
ac9b9c667 [PATCH] Fix handl... |
3001 |
|
f83a275db mm: account for M... |
3002 |
if (vma->vm_flags & VM_MAYSHARE) { |
6bda666a0 [PATCH] hugepages... |
3003 |
int err; |
45c682a68 hugetlb: fix i_bl... |
3004 |
struct inode *inode = mapping->host; |
6bda666a0 [PATCH] hugepages... |
3005 3006 3007 3008 |
err = add_to_page_cache(page, mapping, idx, GFP_KERNEL); if (err) { put_page(page); |
6bda666a0 [PATCH] hugepages... |
3009 3010 3011 3012 |
if (err == -EEXIST) goto retry; goto out; } |
07443a85a mm, hugetlb: retu... |
3013 |
ClearPagePrivate(page); |
45c682a68 hugetlb: fix i_bl... |
3014 3015 |
spin_lock(&inode->i_lock); |
a55164389 hugetlb: modular ... |
3016 |
inode->i_blocks += blocks_per_huge_page(h); |
45c682a68 hugetlb: fix i_bl... |
3017 |
spin_unlock(&inode->i_lock); |
23be7468e hugetlb: fix infi... |
3018 |
} else { |
6bda666a0 [PATCH] hugepages... |
3019 |
lock_page(page); |
0fe6e20b9 hugetlb, rmap: ad... |
3020 3021 3022 3023 |
if (unlikely(anon_vma_prepare(vma))) { ret = VM_FAULT_OOM; goto backout_unlocked; } |
409eb8c26 mm/hugetlb.c: und... |
3024 |
anon_rmap = 1; |
23be7468e hugetlb: fix infi... |
3025 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
3026 |
} else { |
998b4382c hugetlb: fix meta... |
3027 3028 3029 3030 3031 3032 |
/* * 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 ... |
3033 |
ret = VM_FAULT_HWPOISON | |
972dc4de1 hugetlb: add an i... |
3034 |
VM_FAULT_SET_HINDEX(hstate_index(h)); |
998b4382c hugetlb: fix meta... |
3035 3036 |
goto backout_unlocked; } |
6bda666a0 [PATCH] hugepages... |
3037 |
} |
1e8f889b1 [PATCH] Hugetlb: ... |
3038 |
|
57303d801 hugetlbfs: alloca... |
3039 3040 3041 3042 3043 3044 |
/* * If we are going to COW a private mapping later, we examine the * pending reservations for this page now. This will ensure that * any allocations necessary to record that reservation occur outside * the spinlock. */ |
788c7df45 hugetlb: fault fl... |
3045 |
if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) |
2b26736c8 allocate structur... |
3046 3047 3048 3049 |
if (vma_needs_reservation(h, vma, address) < 0) { ret = VM_FAULT_OOM; goto backout_unlocked; } |
57303d801 hugetlbfs: alloca... |
3050 |
|
cb900f412 mm, hugetlb: conv... |
3051 3052 |
ptl = huge_pte_lockptr(h, mm, ptep); spin_lock(ptl); |
a55164389 hugetlb: modular ... |
3053 |
size = i_size_read(mapping->host) >> huge_page_shift(h); |
4c8872659 [PATCH] hugetlb: ... |
3054 3055 |
if (idx >= size) goto backout; |
83c54070e mm: fault feedbac... |
3056 |
ret = 0; |
7f2e9525b hugetlbfs: common... |
3057 |
if (!huge_pte_none(huge_ptep_get(ptep))) |
4c8872659 [PATCH] hugetlb: ... |
3058 |
goto backout; |
07443a85a mm, hugetlb: retu... |
3059 3060 |
if (anon_rmap) { ClearPagePrivate(page); |
409eb8c26 mm/hugetlb.c: und... |
3061 |
hugepage_add_new_anon_rmap(page, vma, address); |
ac7149045 mm: fix 'ERROR: d... |
3062 |
} else |
409eb8c26 mm/hugetlb.c: und... |
3063 |
page_dup_rmap(page); |
1e8f889b1 [PATCH] Hugetlb: ... |
3064 3065 3066 |
new_pte = make_huge_pte(vma, page, ((vma->vm_flags & VM_WRITE) && (vma->vm_flags & VM_SHARED))); set_huge_pte_at(mm, address, ptep, new_pte); |
788c7df45 hugetlb: fault fl... |
3067 |
if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) { |
1e8f889b1 [PATCH] Hugetlb: ... |
3068 |
/* Optimization, do the COW without a second fault */ |
cb900f412 mm, hugetlb: conv... |
3069 |
ret = hugetlb_cow(mm, vma, address, ptep, new_pte, page, ptl); |
1e8f889b1 [PATCH] Hugetlb: ... |
3070 |
} |
cb900f412 mm, hugetlb: conv... |
3071 |
spin_unlock(ptl); |
4c8872659 [PATCH] hugetlb: ... |
3072 3073 |
unlock_page(page); out: |
ac9b9c667 [PATCH] Fix handl... |
3074 |
return ret; |
4c8872659 [PATCH] hugetlb: ... |
3075 3076 |
backout: |
cb900f412 mm, hugetlb: conv... |
3077 |
spin_unlock(ptl); |
2b26736c8 allocate structur... |
3078 |
backout_unlocked: |
4c8872659 [PATCH] hugetlb: ... |
3079 3080 3081 |
unlock_page(page); put_page(page); goto out; |
ac9b9c667 [PATCH] Fix handl... |
3082 |
} |
8382d914e mm, hugetlb: impr... |
3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 |
#ifdef CONFIG_SMP static u32 fault_mutex_hash(struct hstate *h, struct mm_struct *mm, struct vm_area_struct *vma, struct address_space *mapping, pgoff_t idx, unsigned long address) { unsigned long key[2]; u32 hash; if (vma->vm_flags & VM_SHARED) { key[0] = (unsigned long) mapping; key[1] = idx; } else { key[0] = (unsigned long) mm; key[1] = address >> huge_page_shift(h); } hash = jhash2((u32 *)&key, sizeof(key)/sizeof(u32), 0); return hash & (num_fault_mutexes - 1); } #else /* * For uniprocesor systems we always use a single mutex, so just * return 0 and avoid the hashing overhead. */ static u32 fault_mutex_hash(struct hstate *h, struct mm_struct *mm, struct vm_area_struct *vma, struct address_space *mapping, pgoff_t idx, unsigned long address) { return 0; } #endif |
86e5216f8 [PATCH] Hugetlb: ... |
3117 |
int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
788c7df45 hugetlb: fault fl... |
3118 |
unsigned long address, unsigned int flags) |
86e5216f8 [PATCH] Hugetlb: ... |
3119 |
{ |
8382d914e mm, hugetlb: impr... |
3120 |
pte_t *ptep, entry; |
cb900f412 mm, hugetlb: conv... |
3121 |
spinlock_t *ptl; |
1e8f889b1 [PATCH] Hugetlb: ... |
3122 |
int ret; |
8382d914e mm, hugetlb: impr... |
3123 3124 |
u32 hash; pgoff_t idx; |
0fe6e20b9 hugetlb, rmap: ad... |
3125 |
struct page *page = NULL; |
57303d801 hugetlbfs: alloca... |
3126 |
struct page *pagecache_page = NULL; |
a55164389 hugetlb: modular ... |
3127 |
struct hstate *h = hstate_vma(vma); |
8382d914e mm, hugetlb: impr... |
3128 |
struct address_space *mapping; |
0f792cf94 mm/hugetlb: fix g... |
3129 |
int need_wait_lock = 0; |
86e5216f8 [PATCH] Hugetlb: ... |
3130 |
|
1e16a539a mm/hugetlb.c: fix... |
3131 |
address &= huge_page_mask(h); |
fd6a03edd HWPOISON, hugetlb... |
3132 3133 3134 |
ptep = huge_pte_offset(mm, address); if (ptep) { entry = huge_ptep_get(ptep); |
290408d4a hugetlb: hugepage... |
3135 |
if (unlikely(is_hugetlb_entry_migration(entry))) { |
cb900f412 mm, hugetlb: conv... |
3136 |
migration_entry_wait_huge(vma, mm, ptep); |
290408d4a hugetlb: hugepage... |
3137 3138 |
return 0; } else if (unlikely(is_hugetlb_entry_hwpoisoned(entry))) |
32f84528f mm: hugetlb: fix ... |
3139 |
return VM_FAULT_HWPOISON_LARGE | |
972dc4de1 hugetlb: add an i... |
3140 |
VM_FAULT_SET_HINDEX(hstate_index(h)); |
fd6a03edd HWPOISON, hugetlb... |
3141 |
} |
a55164389 hugetlb: modular ... |
3142 |
ptep = huge_pte_alloc(mm, address, huge_page_size(h)); |
86e5216f8 [PATCH] Hugetlb: ... |
3143 3144 |
if (!ptep) return VM_FAULT_OOM; |
8382d914e mm, hugetlb: impr... |
3145 3146 |
mapping = vma->vm_file->f_mapping; idx = vma_hugecache_offset(h, vma, address); |
3935baa9b [PATCH] hugepage:... |
3147 3148 3149 3150 3151 |
/* * Serialize hugepage allocation and instantiation, so that we don't * get spurious allocation failures if two CPUs race to instantiate * the same page in the page cache. */ |
8382d914e mm, hugetlb: impr... |
3152 3153 |
hash = fault_mutex_hash(h, mm, vma, mapping, idx, address); mutex_lock(&htlb_fault_mutex_table[hash]); |
7f2e9525b hugetlbfs: common... |
3154 3155 |
entry = huge_ptep_get(ptep); if (huge_pte_none(entry)) { |
8382d914e mm, hugetlb: impr... |
3156 |
ret = hugetlb_no_page(mm, vma, mapping, idx, address, ptep, flags); |
b4d1d99fd hugetlb: handle u... |
3157 |
goto out_mutex; |
3935baa9b [PATCH] hugepage:... |
3158 |
} |
86e5216f8 [PATCH] Hugetlb: ... |
3159 |
|
83c54070e mm: fault feedbac... |
3160 |
ret = 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
3161 |
|
57303d801 hugetlbfs: alloca... |
3162 |
/* |
0f792cf94 mm/hugetlb: fix g... |
3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 |
* 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... |
3173 3174 3175 3176 3177 3178 3179 |
* 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... |
3180 |
if ((flags & FAULT_FLAG_WRITE) && !huge_pte_write(entry)) { |
2b26736c8 allocate structur... |
3181 3182 |
if (vma_needs_reservation(h, vma, address) < 0) { ret = VM_FAULT_OOM; |
b4d1d99fd hugetlb: handle u... |
3183 |
goto out_mutex; |
2b26736c8 allocate structur... |
3184 |
} |
57303d801 hugetlbfs: alloca... |
3185 |
|
f83a275db mm: account for M... |
3186 |
if (!(vma->vm_flags & VM_MAYSHARE)) |
57303d801 hugetlbfs: alloca... |
3187 3188 3189 |
pagecache_page = hugetlbfs_pagecache_page(h, vma, address); } |
0f792cf94 mm/hugetlb: fix g... |
3190 3191 3192 3193 3194 |
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... |
3195 3196 3197 3198 |
/* * 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... |
3199 3200 3201 |
*/ page = pte_page(entry); if (page != pagecache_page) |
0f792cf94 mm/hugetlb: fix g... |
3202 3203 3204 3205 |
if (!trylock_page(page)) { need_wait_lock = 1; goto out_ptl; } |
b4d1d99fd hugetlb: handle u... |
3206 |
|
0f792cf94 mm/hugetlb: fix g... |
3207 |
get_page(page); |
b4d1d99fd hugetlb: handle u... |
3208 |
|
788c7df45 hugetlb: fault fl... |
3209 |
if (flags & FAULT_FLAG_WRITE) { |
106c992a5 mm/hugetlb: add m... |
3210 |
if (!huge_pte_write(entry)) { |
57303d801 hugetlbfs: alloca... |
3211 |
ret = hugetlb_cow(mm, vma, address, ptep, entry, |
cb900f412 mm, hugetlb: conv... |
3212 |
pagecache_page, ptl); |
0f792cf94 mm/hugetlb: fix g... |
3213 |
goto out_put_page; |
b4d1d99fd hugetlb: handle u... |
3214 |
} |
106c992a5 mm/hugetlb: add m... |
3215 |
entry = huge_pte_mkdirty(entry); |
b4d1d99fd hugetlb: handle u... |
3216 3217 |
} entry = pte_mkyoung(entry); |
788c7df45 hugetlb: fault fl... |
3218 3219 |
if (huge_ptep_set_access_flags(vma, address, ptep, entry, flags & FAULT_FLAG_WRITE)) |
4b3073e1c MM: Pass a PTE po... |
3220 |
update_mmu_cache(vma, address, ptep); |
0f792cf94 mm/hugetlb: fix g... |
3221 3222 3223 3224 |
out_put_page: if (page != pagecache_page) unlock_page(page); put_page(page); |
cb900f412 mm, hugetlb: conv... |
3225 3226 |
out_ptl: spin_unlock(ptl); |
57303d801 hugetlbfs: alloca... |
3227 3228 3229 3230 3231 |
if (pagecache_page) { unlock_page(pagecache_page); put_page(pagecache_page); } |
b4d1d99fd hugetlb: handle u... |
3232 |
out_mutex: |
8382d914e mm, hugetlb: impr... |
3233 |
mutex_unlock(&htlb_fault_mutex_table[hash]); |
0f792cf94 mm/hugetlb: fix g... |
3234 3235 3236 3237 3238 3239 3240 3241 3242 |
/* * 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: ... |
3243 |
return ret; |
86e5216f8 [PATCH] Hugetlb: ... |
3244 |
} |
28a35716d mm: use long type... |
3245 3246 3247 3248 |
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 ... |
3249 |
{ |
d5d4b0aa4 [PATCH] optimize ... |
3250 3251 |
unsigned long pfn_offset; unsigned long vaddr = *position; |
28a35716d mm: use long type... |
3252 |
unsigned long remainder = *nr_pages; |
a55164389 hugetlb: modular ... |
3253 |
struct hstate *h = hstate_vma(vma); |
63551ae0f [PATCH] Hugepage ... |
3254 |
|
63551ae0f [PATCH] Hugepage ... |
3255 |
while (vaddr < vma->vm_end && remainder) { |
4c8872659 [PATCH] hugetlb: ... |
3256 |
pte_t *pte; |
cb900f412 mm, hugetlb: conv... |
3257 |
spinlock_t *ptl = NULL; |
2a15efc95 mm: follow_hugetl... |
3258 |
int absent; |
4c8872659 [PATCH] hugetlb: ... |
3259 |
struct page *page; |
63551ae0f [PATCH] Hugepage ... |
3260 |
|
4c8872659 [PATCH] hugetlb: ... |
3261 |
/* |
02057967b mm, hugetlb: abor... |
3262 3263 3264 3265 3266 3267 3268 3269 3270 |
* 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: ... |
3271 |
* Some archs (sparc64, sh*) have multiple pte_ts to |
2a15efc95 mm: follow_hugetl... |
3272 |
* each hugepage. We have to make sure we get the |
4c8872659 [PATCH] hugetlb: ... |
3273 |
* first, for the page indexing below to work. |
cb900f412 mm, hugetlb: conv... |
3274 3275 |
* * Note that page table lock is not held when pte is null. |
4c8872659 [PATCH] hugetlb: ... |
3276 |
*/ |
a55164389 hugetlb: modular ... |
3277 |
pte = huge_pte_offset(mm, vaddr & huge_page_mask(h)); |
cb900f412 mm, hugetlb: conv... |
3278 3279 |
if (pte) ptl = huge_pte_lock(h, mm, pte); |
2a15efc95 mm: follow_hugetl... |
3280 3281 3282 3283 |
absent = !pte || huge_pte_none(huge_ptep_get(pte)); /* * When coredumping, it suits get_dump_page if we just return |
3ae77f43b mm: hugetlbfs_pag... |
3284 3285 3286 3287 |
* 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... |
3288 |
*/ |
3ae77f43b mm: hugetlbfs_pag... |
3289 3290 |
if (absent && (flags & FOLL_DUMP) && !hugetlbfs_pagecache_present(h, vma, vaddr)) { |
cb900f412 mm, hugetlb: conv... |
3291 3292 |
if (pte) spin_unlock(ptl); |
2a15efc95 mm: follow_hugetl... |
3293 3294 3295 |
remainder = 0; break; } |
63551ae0f [PATCH] Hugepage ... |
3296 |
|
9cc3a5bd4 hugetlbfs: add sw... |
3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 |
/* * 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... |
3308 3309 |
((flags & FOLL_WRITE) && !huge_pte_write(huge_ptep_get(pte)))) { |
4c8872659 [PATCH] hugetlb: ... |
3310 |
int ret; |
63551ae0f [PATCH] Hugepage ... |
3311 |
|
cb900f412 mm, hugetlb: conv... |
3312 3313 |
if (pte) spin_unlock(ptl); |
2a15efc95 mm: follow_hugetl... |
3314 3315 |
ret = hugetlb_fault(mm, vma, vaddr, (flags & FOLL_WRITE) ? FAULT_FLAG_WRITE : 0); |
a89182c76 Fix VM_FAULT flag... |
3316 |
if (!(ret & VM_FAULT_ERROR)) |
4c8872659 [PATCH] hugetlb: ... |
3317 |
continue; |
63551ae0f [PATCH] Hugepage ... |
3318 |
|
4c8872659 [PATCH] hugetlb: ... |
3319 |
remainder = 0; |
4c8872659 [PATCH] hugetlb: ... |
3320 3321 |
break; } |
a55164389 hugetlb: modular ... |
3322 |
pfn_offset = (vaddr & ~huge_page_mask(h)) >> PAGE_SHIFT; |
7f2e9525b hugetlbfs: common... |
3323 |
page = pte_page(huge_ptep_get(pte)); |
d5d4b0aa4 [PATCH] optimize ... |
3324 |
same_page: |
d6692183a [PATCH] fix extra... |
3325 |
if (pages) { |
2a15efc95 mm: follow_hugetl... |
3326 |
pages[i] = mem_map_offset(page, pfn_offset); |
a0368d4e4 mm: hugetlb: use ... |
3327 |
get_page_foll(pages[i]); |
d6692183a [PATCH] fix extra... |
3328 |
} |
63551ae0f [PATCH] Hugepage ... |
3329 3330 3331 3332 3333 |
if (vmas) vmas[i] = vma; vaddr += PAGE_SIZE; |
d5d4b0aa4 [PATCH] optimize ... |
3334 |
++pfn_offset; |
63551ae0f [PATCH] Hugepage ... |
3335 3336 |
--remainder; ++i; |
d5d4b0aa4 [PATCH] optimize ... |
3337 |
if (vaddr < vma->vm_end && remainder && |
a55164389 hugetlb: modular ... |
3338 |
pfn_offset < pages_per_huge_page(h)) { |
d5d4b0aa4 [PATCH] optimize ... |
3339 3340 3341 3342 3343 3344 |
/* * We use pfn_offset to avoid touching the pageframes * of this compound page. */ goto same_page; } |
cb900f412 mm, hugetlb: conv... |
3345 |
spin_unlock(ptl); |
63551ae0f [PATCH] Hugepage ... |
3346 |
} |
28a35716d mm: use long type... |
3347 |
*nr_pages = remainder; |
63551ae0f [PATCH] Hugepage ... |
3348 |
*position = vaddr; |
2a15efc95 mm: follow_hugetl... |
3349 |
return i ? i : -EFAULT; |
63551ae0f [PATCH] Hugepage ... |
3350 |
} |
8f860591f [PATCH] Enable mp... |
3351 |
|
7da4d641c mm: Count the num... |
3352 |
unsigned long hugetlb_change_protection(struct vm_area_struct *vma, |
8f860591f [PATCH] Enable mp... |
3353 3354 3355 3356 3357 3358 |
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 ... |
3359 |
struct hstate *h = hstate_vma(vma); |
7da4d641c mm: Count the num... |
3360 |
unsigned long pages = 0; |
8f860591f [PATCH] Enable mp... |
3361 3362 3363 |
BUG_ON(address >= end); flush_cache_range(vma, address, end); |
a5338093b mm: move mmu noti... |
3364 |
mmu_notifier_invalidate_range_start(mm, start, end); |
83cde9e8b mm: use new helpe... |
3365 |
i_mmap_lock_write(vma->vm_file->f_mapping); |
a55164389 hugetlb: modular ... |
3366 |
for (; address < end; address += huge_page_size(h)) { |
cb900f412 mm, hugetlb: conv... |
3367 |
spinlock_t *ptl; |
8f860591f [PATCH] Enable mp... |
3368 3369 3370 |
ptep = huge_pte_offset(mm, address); if (!ptep) continue; |
cb900f412 mm, hugetlb: conv... |
3371 |
ptl = huge_pte_lock(h, mm, ptep); |
7da4d641c mm: Count the num... |
3372 3373 |
if (huge_pmd_unshare(mm, &address, ptep)) { pages++; |
cb900f412 mm, hugetlb: conv... |
3374 |
spin_unlock(ptl); |
39dde65c9 [PATCH] shared pa... |
3375 |
continue; |
7da4d641c mm: Count the num... |
3376 |
} |
a8bda28d8 mm/hugetlb: add m... |
3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 |
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... |
3397 |
pte = huge_ptep_get_and_clear(mm, address, ptep); |
106c992a5 mm/hugetlb: add m... |
3398 |
pte = pte_mkhuge(huge_pte_modify(pte, newprot)); |
be7517d6a mm/hugetlb: set P... |
3399 |
pte = arch_make_huge_pte(pte, vma, NULL, 0); |
8f860591f [PATCH] Enable mp... |
3400 |
set_huge_pte_at(mm, address, ptep, pte); |
7da4d641c mm: Count the num... |
3401 |
pages++; |
8f860591f [PATCH] Enable mp... |
3402 |
} |
cb900f412 mm, hugetlb: conv... |
3403 |
spin_unlock(ptl); |
8f860591f [PATCH] Enable mp... |
3404 |
} |
d833352a4 mm: hugetlbfs: cl... |
3405 |
/* |
c8c06efa8 mm: convert i_mma... |
3406 |
* Must flush TLB before releasing i_mmap_rwsem: x86's huge_pmd_unshare |
d833352a4 mm: hugetlbfs: cl... |
3407 |
* may have cleared our pud entry and done put_page on the page table: |
c8c06efa8 mm: convert i_mma... |
3408 |
* once we release i_mmap_rwsem, another task can do the final put_page |
d833352a4 mm: hugetlbfs: cl... |
3409 3410 |
* and that page table be reused and filled with junk. */ |
8f860591f [PATCH] Enable mp... |
3411 |
flush_tlb_range(vma, start, end); |
34ee645e8 mmu_notifier: cal... |
3412 |
mmu_notifier_invalidate_range(mm, start, end); |
83cde9e8b mm: use new helpe... |
3413 |
i_mmap_unlock_write(vma->vm_file->f_mapping); |
a5338093b mm: move mmu noti... |
3414 |
mmu_notifier_invalidate_range_end(mm, start, end); |
7da4d641c mm: Count the num... |
3415 3416 |
return pages << h->order; |
8f860591f [PATCH] Enable mp... |
3417 |
} |
a1e78772d hugetlb: reserve ... |
3418 3419 |
int hugetlb_reserve_pages(struct inode *inode, long from, long to, |
5a6fe1259 Do not account fo... |
3420 |
struct vm_area_struct *vma, |
ca16d140a mm: don't access ... |
3421 |
vm_flags_t vm_flags) |
e4e574b76 hugetlb: Try to g... |
3422 |
{ |
17c9d12e1 Do not account fo... |
3423 |
long ret, chg; |
a55164389 hugetlb: modular ... |
3424 |
struct hstate *h = hstate_inode(inode); |
90481622d hugepages: fix us... |
3425 |
struct hugepage_subpool *spool = subpool_inode(inode); |
9119a41e9 mm, hugetlb: unif... |
3426 |
struct resv_map *resv_map; |
1c5ecae3a hugetlbfs: add mi... |
3427 |
long gbl_reserve; |
e4e574b76 hugetlb: Try to g... |
3428 |
|
a1e78772d hugetlb: reserve ... |
3429 |
/* |
17c9d12e1 Do not account fo... |
3430 3431 |
* 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... |
3432 |
* without using reserves |
17c9d12e1 Do not account fo... |
3433 |
*/ |
ca16d140a mm: don't access ... |
3434 |
if (vm_flags & VM_NORESERVE) |
17c9d12e1 Do not account fo... |
3435 3436 3437 |
return 0; /* |
a1e78772d hugetlb: reserve ... |
3438 3439 3440 3441 3442 |
* 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... |
3443 |
if (!vma || vma->vm_flags & VM_MAYSHARE) { |
4e35f4838 mm, hugetlb: use ... |
3444 |
resv_map = inode_resv_map(inode); |
9119a41e9 mm, hugetlb: unif... |
3445 |
|
1406ec9ba mm, hugetlb: impr... |
3446 |
chg = region_chg(resv_map, from, to); |
9119a41e9 mm, hugetlb: unif... |
3447 3448 3449 |
} else { resv_map = resv_map_alloc(); |
17c9d12e1 Do not account fo... |
3450 3451 |
if (!resv_map) return -ENOMEM; |
a1e78772d hugetlb: reserve ... |
3452 |
chg = to - from; |
84afd99b8 hugetlb reservati... |
3453 |
|
17c9d12e1 Do not account fo... |
3454 3455 3456 |
set_vma_resv_map(vma, resv_map); set_vma_resv_flags(vma, HPAGE_RESV_OWNER); } |
c50ac0508 hugetlb: fix resv... |
3457 3458 3459 3460 |
if (chg < 0) { ret = chg; goto out_err; } |
8a6301127 pretend cpuset ha... |
3461 |
|
1c5ecae3a hugetlbfs: add mi... |
3462 3463 3464 3465 3466 3467 3468 |
/* * 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... |
3469 3470 3471 |
ret = -ENOSPC; goto out_err; } |
5a6fe1259 Do not account fo... |
3472 3473 |
/* |
17c9d12e1 Do not account fo... |
3474 |
* Check enough hugepages are available for the reservation. |
90481622d hugepages: fix us... |
3475 |
* Hand the pages back to the subpool if there are not |
5a6fe1259 Do not account fo... |
3476 |
*/ |
1c5ecae3a hugetlbfs: add mi... |
3477 |
ret = hugetlb_acct_memory(h, gbl_reserve); |
68842c9b9 hugetlbfs: fix qu... |
3478 |
if (ret < 0) { |
1c5ecae3a hugetlbfs: add mi... |
3479 3480 |
/* put back original number of pages, chg */ (void)hugepage_subpool_put_pages(spool, chg); |
c50ac0508 hugetlb: fix resv... |
3481 |
goto out_err; |
68842c9b9 hugetlbfs: fix qu... |
3482 |
} |
17c9d12e1 Do not account fo... |
3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 |
/* * Account for the reservations made. Shared mappings record regions * that have reservations as they are shared by multiple VMAs. * When the last VMA disappears, the region map says how much * the reservation was and the page cache tells how much of * the reservation was consumed. Private mappings are per-VMA and * only the consumed reservations are tracked. When the VMA * disappears, the original reservation is the VMA size and the * consumed reservations are stored in the map. Hence, nothing * else has to be done for private mappings here */ |
f83a275db mm: account for M... |
3495 |
if (!vma || vma->vm_flags & VM_MAYSHARE) |
1406ec9ba mm, hugetlb: impr... |
3496 |
region_add(resv_map, from, to); |
a43a8c39b [PATCH] tightenin... |
3497 |
return 0; |
c50ac0508 hugetlb: fix resv... |
3498 |
out_err: |
f031dd274 mm, hugetlb: remo... |
3499 3500 |
if (vma && is_vma_resv_set(vma, HPAGE_RESV_OWNER)) kref_put(&resv_map->refs, resv_map_release); |
c50ac0508 hugetlb: fix resv... |
3501 |
return ret; |
a43a8c39b [PATCH] tightenin... |
3502 3503 3504 3505 |
} void hugetlb_unreserve_pages(struct inode *inode, long offset, long freed) { |
a55164389 hugetlb: modular ... |
3506 |
struct hstate *h = hstate_inode(inode); |
4e35f4838 mm, hugetlb: use ... |
3507 |
struct resv_map *resv_map = inode_resv_map(inode); |
9119a41e9 mm, hugetlb: unif... |
3508 |
long chg = 0; |
90481622d hugepages: fix us... |
3509 |
struct hugepage_subpool *spool = subpool_inode(inode); |
1c5ecae3a hugetlbfs: add mi... |
3510 |
long gbl_reserve; |
45c682a68 hugetlb: fix i_bl... |
3511 |
|
9119a41e9 mm, hugetlb: unif... |
3512 |
if (resv_map) |
1406ec9ba mm, hugetlb: impr... |
3513 |
chg = region_truncate(resv_map, offset); |
45c682a68 hugetlb: fix i_bl... |
3514 |
spin_lock(&inode->i_lock); |
e4c6f8bed hugetlbfs: fix i_... |
3515 |
inode->i_blocks -= (blocks_per_huge_page(h) * freed); |
45c682a68 hugetlb: fix i_bl... |
3516 |
spin_unlock(&inode->i_lock); |
1c5ecae3a hugetlbfs: add mi... |
3517 3518 3519 3520 3521 3522 |
/* * 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); |
a43a8c39b [PATCH] tightenin... |
3523 |
} |
93f70f900 HWPOISON, hugetlb... |
3524 |
|
3212b535f mm: hugetlb: Copy... |
3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 |
#ifdef CONFIG_ARCH_WANT_HUGE_PMD_SHARE static unsigned long page_table_shareable(struct vm_area_struct *svma, struct vm_area_struct *vma, unsigned long addr, pgoff_t idx) { unsigned long saddr = ((idx - svma->vm_pgoff) << PAGE_SHIFT) + svma->vm_start; unsigned long sbase = saddr & PUD_MASK; unsigned long s_end = sbase + PUD_SIZE; /* Allow segments to share if only one is marked locked */ unsigned long vm_flags = vma->vm_flags & ~VM_LOCKED; unsigned long svm_flags = svma->vm_flags & ~VM_LOCKED; /* * match the virtual addresses, permission and the alignment of the * page table page. */ if (pmd_index(addr) != pmd_index(saddr) || vm_flags != svm_flags || sbase < svma->vm_start || svma->vm_end < s_end) return 0; return saddr; } static int vma_shareable(struct vm_area_struct *vma, unsigned long addr) { unsigned long base = addr & PUD_MASK; unsigned long end = base + PUD_SIZE; /* * check on proper vm_flags and page table alignment */ if (vma->vm_flags & VM_MAYSHARE && vma->vm_start <= base && end <= vma->vm_end) return 1; return 0; } /* * Search for a shareable pmd page for hugetlb. In any case calls pmd_alloc() * and returns the corresponding pte. While this is not necessary for the * !shared pmd case because we can allocate the pmd later as well, it makes the * code much cleaner. pmd allocation is essential for the shared case because |
c8c06efa8 mm: convert i_mma... |
3570 |
* pud has to be populated inside the same i_mmap_rwsem section - otherwise |
3212b535f mm: hugetlb: Copy... |
3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 |
* 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... |
3584 |
spinlock_t *ptl; |
3212b535f mm: hugetlb: Copy... |
3585 3586 3587 |
if (!vma_shareable(vma, addr)) return (pte_t *)pmd_alloc(mm, pud, addr); |
83cde9e8b mm: use new helpe... |
3588 |
i_mmap_lock_write(mapping); |
3212b535f mm: hugetlb: Copy... |
3589 3590 3591 3592 3593 3594 3595 3596 |
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) { |
dc6c9a35b mm: account pmd p... |
3597 |
mm_inc_nr_pmds(mm); |
3212b535f mm: hugetlb: Copy... |
3598 3599 3600 3601 3602 3603 3604 3605 |
get_page(virt_to_page(spte)); break; } } } if (!spte) goto out; |
cb900f412 mm, hugetlb: conv... |
3606 3607 |
ptl = huge_pte_lockptr(hstate_vma(vma), mm, spte); spin_lock(ptl); |
dc6c9a35b mm: account pmd p... |
3608 |
if (pud_none(*pud)) { |
3212b535f mm: hugetlb: Copy... |
3609 3610 |
pud_populate(mm, pud, (pmd_t *)((unsigned long)spte & PAGE_MASK)); |
dc6c9a35b mm: account pmd p... |
3611 |
} else { |
3212b535f mm: hugetlb: Copy... |
3612 |
put_page(virt_to_page(spte)); |
dc6c9a35b mm: account pmd p... |
3613 3614 |
mm_inc_nr_pmds(mm); } |
cb900f412 mm, hugetlb: conv... |
3615 |
spin_unlock(ptl); |
3212b535f mm: hugetlb: Copy... |
3616 3617 |
out: pte = (pte_t *)pmd_alloc(mm, pud, addr); |
83cde9e8b mm: use new helpe... |
3618 |
i_mmap_unlock_write(mapping); |
3212b535f mm: hugetlb: Copy... |
3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 |
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... |
3629 |
* called with page table lock held. |
3212b535f mm: hugetlb: Copy... |
3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 |
* * 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... |
3645 |
mm_dec_nr_pmds(mm); |
3212b535f mm: hugetlb: Copy... |
3646 3647 3648 |
*addr = ALIGN(*addr, HPAGE_SIZE * PTRS_PER_PTE) - HPAGE_SIZE; return 1; } |
9e5fc74c3 mm: hugetlb: Copy... |
3649 3650 3651 3652 3653 3654 3655 |
#define want_pmd_share() (1) #else /* !CONFIG_ARCH_WANT_HUGE_PMD_SHARE */ pte_t *huge_pmd_share(struct mm_struct *mm, unsigned long addr, pud_t *pud) { return NULL; } #define want_pmd_share() (0) |
3212b535f mm: hugetlb: Copy... |
3656 |
#endif /* CONFIG_ARCH_WANT_HUGE_PMD_SHARE */ |
9e5fc74c3 mm: hugetlb: Copy... |
3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 |
#ifdef CONFIG_ARCH_WANT_GENERAL_HUGETLB pte_t *huge_pte_alloc(struct mm_struct *mm, unsigned long addr, unsigned long sz) { pgd_t *pgd; pud_t *pud; pte_t *pte = NULL; pgd = pgd_offset(mm, addr); pud = pud_alloc(mm, pgd, addr); if (pud) { if (sz == PUD_SIZE) { pte = (pte_t *)pud; } else { BUG_ON(sz != PMD_SIZE); if (want_pmd_share() && pud_none(*pud)) pte = huge_pmd_share(mm, addr, pud); else pte = (pte_t *)pmd_alloc(mm, pud, addr); } } BUG_ON(pte && !pte_none(*pte) && !pte_huge(*pte)); return pte; } pte_t *huge_pte_offset(struct mm_struct *mm, unsigned long addr) { pgd_t *pgd; pud_t *pud; pmd_t *pmd = NULL; pgd = pgd_offset(mm, addr); if (pgd_present(*pgd)) { pud = pud_offset(pgd, addr); if (pud_present(*pud)) { if (pud_huge(*pud)) return (pte_t *)pud; pmd = pmd_offset(pud, addr); } } return (pte_t *) pmd; } |
61f77eda9 mm/hugetlb: reduc... |
3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 |
#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... |
3714 |
follow_huge_pmd(struct mm_struct *mm, unsigned long address, |
e66f17ff7 mm/hugetlb: take ... |
3715 |
pmd_t *pmd, int flags) |
9e5fc74c3 mm: hugetlb: Copy... |
3716 |
{ |
e66f17ff7 mm/hugetlb: take ... |
3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 |
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... |
3729 |
page = pmd_page(*pmd) + ((address & ~PMD_MASK) >> PAGE_SHIFT); |
e66f17ff7 mm/hugetlb: take ... |
3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 |
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... |
3745 3746 |
return page; } |
61f77eda9 mm/hugetlb: reduc... |
3747 |
struct page * __weak |
9e5fc74c3 mm: hugetlb: Copy... |
3748 |
follow_huge_pud(struct mm_struct *mm, unsigned long address, |
e66f17ff7 mm/hugetlb: take ... |
3749 |
pud_t *pud, int flags) |
9e5fc74c3 mm: hugetlb: Copy... |
3750 |
{ |
e66f17ff7 mm/hugetlb: take ... |
3751 3752 |
if (flags & FOLL_GET) return NULL; |
9e5fc74c3 mm: hugetlb: Copy... |
3753 |
|
e66f17ff7 mm/hugetlb: take ... |
3754 |
return pte_page(*(pte_t *)pud) + ((address & ~PUD_MASK) >> PAGE_SHIFT); |
9e5fc74c3 mm: hugetlb: Copy... |
3755 |
} |
d5bd91069 hugepage: move is... |
3756 |
#ifdef CONFIG_MEMORY_FAILURE |
93f70f900 HWPOISON, hugetlb... |
3757 3758 3759 3760 |
/* * This function is called from memory failure code. * Assume the caller holds page lock of the head page. */ |
6de2b1aab HWPOISON, hugetlb... |
3761 |
int dequeue_hwpoisoned_huge_page(struct page *hpage) |
93f70f900 HWPOISON, hugetlb... |
3762 3763 3764 |
{ struct hstate *h = page_hstate(hpage); int nid = page_to_nid(hpage); |
6de2b1aab HWPOISON, hugetlb... |
3765 |
int ret = -EBUSY; |
93f70f900 HWPOISON, hugetlb... |
3766 3767 |
spin_lock(&hugetlb_lock); |
7e1f049ef mm: hugetlb: clea... |
3768 3769 3770 3771 3772 |
/* * 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... |
3773 3774 3775 3776 3777 3778 3779 |
/* * 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... |
3780 |
set_page_refcounted(hpage); |
6de2b1aab HWPOISON, hugetlb... |
3781 3782 3783 3784 |
h->free_huge_pages--; h->free_huge_pages_node[nid]--; ret = 0; } |
93f70f900 HWPOISON, hugetlb... |
3785 |
spin_unlock(&hugetlb_lock); |
6de2b1aab HWPOISON, hugetlb... |
3786 |
return ret; |
93f70f900 HWPOISON, hugetlb... |
3787 |
} |
6de2b1aab HWPOISON, hugetlb... |
3788 |
#endif |
31caf665e mm: migrate: make... |
3789 3790 3791 |
bool isolate_huge_page(struct page *page, struct list_head *list) { |
bcc542223 mm: hugetlb: intr... |
3792 |
bool ret = true; |
309381fea mm: dump page whe... |
3793 |
VM_BUG_ON_PAGE(!PageHead(page), page); |
31caf665e mm: migrate: make... |
3794 |
spin_lock(&hugetlb_lock); |
bcc542223 mm: hugetlb: intr... |
3795 3796 3797 3798 3799 |
if (!page_huge_active(page) || !get_page_unless_zero(page)) { ret = false; goto unlock; } clear_page_huge_active(page); |
31caf665e mm: migrate: make... |
3800 |
list_move_tail(&page->lru, list); |
bcc542223 mm: hugetlb: intr... |
3801 |
unlock: |
31caf665e mm: migrate: make... |
3802 |
spin_unlock(&hugetlb_lock); |
bcc542223 mm: hugetlb: intr... |
3803 |
return ret; |
31caf665e mm: migrate: make... |
3804 3805 3806 3807 |
} void putback_active_hugepage(struct page *page) { |
309381fea mm: dump page whe... |
3808 |
VM_BUG_ON_PAGE(!PageHead(page), page); |
31caf665e mm: migrate: make... |
3809 |
spin_lock(&hugetlb_lock); |
bcc542223 mm: hugetlb: intr... |
3810 |
set_page_huge_active(page); |
31caf665e mm: migrate: make... |
3811 3812 3813 3814 |
list_move_tail(&page->lru, &(page_hstate(page))->hugepage_activelist); spin_unlock(&hugetlb_lock); put_page(page); } |