Blame view
mm/hugetlb.c
76.9 KB
1da177e4c Linux-2.6.12-rc2 |
1 2 3 4 |
/* * Generic hugetlb support. * (C) William Irwin, April 2004 */ |
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> |
aea47ff36 [PATCH] mm: make ... |
16 |
#include <linux/cpuset.h> |
3935baa9b [PATCH] hugepage:... |
17 |
#include <linux/mutex.h> |
aa888a749 hugetlb: support ... |
18 |
#include <linux/bootmem.h> |
a34378701 hugetlb: new sysf... |
19 |
#include <linux/sysfs.h> |
5a0e3ad6a include cleanup: ... |
20 |
#include <linux/slab.h> |
0fe6e20b9 hugetlb, rmap: ad... |
21 |
#include <linux/rmap.h> |
fd6a03edd HWPOISON, hugetlb... |
22 23 |
#include <linux/swap.h> #include <linux/swapops.h> |
d6606683a Revert duplicate ... |
24 |
|
63551ae0f [PATCH] Hugepage ... |
25 26 |
#include <asm/page.h> #include <asm/pgtable.h> |
78a34ae29 mm/hugetlb.c must... |
27 |
#include <asm/io.h> |
63551ae0f [PATCH] Hugepage ... |
28 29 |
#include <linux/hugetlb.h> |
9a3052306 hugetlb: add per ... |
30 |
#include <linux/node.h> |
7835e98b2 [PATCH] remove se... |
31 |
#include "internal.h" |
1da177e4c Linux-2.6.12-rc2 |
32 33 |
const unsigned long hugetlb_zero = 0, hugetlb_infinity = ~0UL; |
396faf030 Allow huge page a... |
34 35 |
static gfp_t htlb_alloc_mask = GFP_HIGHUSER; unsigned long hugepages_treat_as_movable; |
a55164389 hugetlb: modular ... |
36 |
|
e5ff21594 hugetlb: multiple... |
37 38 39 |
static int max_hstate; unsigned int default_hstate_idx; struct hstate hstates[HUGE_MAX_HSTATE]; |
53ba51d21 hugetlb: allow ar... |
40 |
__initdata LIST_HEAD(huge_boot_pages); |
e5ff21594 hugetlb: multiple... |
41 42 43 |
/* for command line parsing */ static struct hstate * __initdata parsed_hstate; static unsigned long __initdata default_hstate_max_huge_pages; |
e11bfbfcb hugetlb: override... |
44 |
static unsigned long __initdata default_hstate_size; |
e5ff21594 hugetlb: multiple... |
45 46 47 |
#define for_each_hstate(h) \ for ((h) = hstates; (h) < &hstates[max_hstate]; (h)++) |
396faf030 Allow huge page a... |
48 |
|
3935baa9b [PATCH] hugepage:... |
49 50 51 52 |
/* * Protects updates to hugepage_freelists, nr_huge_pages, and free_huge_pages */ static DEFINE_SPINLOCK(hugetlb_lock); |
0bd0f9fb1 [PATCH] hugetlb: ... |
53 |
|
e7c4b0bfd huge page private... |
54 |
/* |
968229048 hugetlb: move res... |
55 56 |
* Region tracking -- allows tracking of reservations and instantiated pages * across the pages in a mapping. |
84afd99b8 hugetlb reservati... |
57 58 59 60 61 62 63 64 65 66 |
* * The region data structures are protected by a combination of the mmap_sem * and the hugetlb_instantion_mutex. To access or modify a region the caller * must either hold the mmap_sem for write, or the mmap_sem for read and * the hugetlb_instantiation mutex: * * down_write(&mm->mmap_sem); * or * down_read(&mm->mmap_sem); * mutex_lock(&hugetlb_instantiation_mutex); |
968229048 hugetlb: move res... |
67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 |
*/ struct file_region { struct list_head link; long from; long to; }; static long region_add(struct list_head *head, long f, long t) { struct file_region *rg, *nrg, *trg; /* Locate the region we are either in or before. */ list_for_each_entry(rg, head, link) if (f <= rg->to) break; /* Round our left edge to the current segment if it encloses us. */ if (f > rg->from) f = rg->from; /* Check for and consume any regions we now overlap with. */ nrg = rg; list_for_each_entry_safe(rg, trg, rg->link.prev, link) { if (&rg->link == head) break; if (rg->from > t) break; /* If this area reaches higher then extend our area to * include it completely. If this is not the first area * which we intend to reuse, free it. */ if (rg->to > t) t = rg->to; if (rg != nrg) { list_del(&rg->link); kfree(rg); } } nrg->from = f; nrg->to = t; return 0; } static long region_chg(struct list_head *head, long f, long t) { struct file_region *rg, *nrg; long chg = 0; /* Locate the region we are before or in. */ list_for_each_entry(rg, head, link) if (f <= rg->to) break; /* If we are below the current region then a new region is required. * Subtle, allocate a new region at the position but make it zero * size such that we can guarantee to record the reservation. */ if (&rg->link == head || t < rg->from) { nrg = kmalloc(sizeof(*nrg), GFP_KERNEL); if (!nrg) return -ENOMEM; nrg->from = f; nrg->to = f; INIT_LIST_HEAD(&nrg->link); list_add(&nrg->link, rg->link.prev); return t - f; } /* Round our left edge to the current segment if it encloses us. */ if (f > rg->from) f = rg->from; chg = t - f; /* Check for and consume any regions we now overlap with. */ list_for_each_entry(rg, rg->link.prev, link) { if (&rg->link == head) break; if (rg->from > t) return chg; /* We overlap with this area, if it extends futher than * us then we must extend ourselves. Account for its * existing reservation. */ if (rg->to > t) { chg += rg->to - t; t = rg->to; } chg -= rg->to - rg->from; } return chg; } static long region_truncate(struct list_head *head, long end) { struct file_region *rg, *trg; long chg = 0; /* Locate the region we are either in or before. */ list_for_each_entry(rg, head, link) if (end <= rg->to) break; if (&rg->link == head) return 0; /* If we are in the middle of a region then adjust it. */ if (end > rg->from) { chg = rg->to - end; rg->to = end; rg = list_entry(rg->link.next, typeof(*rg), link); } /* Drop any remaining regions. */ list_for_each_entry_safe(rg, trg, rg->link.prev, link) { if (&rg->link == head) break; chg += rg->to - rg->from; list_del(&rg->link); kfree(rg); } return chg; } |
84afd99b8 hugetlb reservati... |
188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 |
static long region_count(struct list_head *head, long f, long t) { struct file_region *rg; long chg = 0; /* Locate each segment we overlap with, and count that overlap. */ list_for_each_entry(rg, head, link) { int seg_from; int seg_to; if (rg->to <= f) continue; if (rg->from >= t) break; seg_from = max(rg->from, f); seg_to = min(rg->to, t); chg += seg_to - seg_from; } return chg; } |
968229048 hugetlb: move res... |
211 |
/* |
e7c4b0bfd huge page private... |
212 |
* Convert the address within this vma to the page offset within |
e7c4b0bfd huge page private... |
213 214 |
* the mapping, in pagecache page units; huge pages here. */ |
a55164389 hugetlb: modular ... |
215 216 |
static pgoff_t vma_hugecache_offset(struct hstate *h, struct vm_area_struct *vma, unsigned long address) |
e7c4b0bfd huge page private... |
217 |
{ |
a55164389 hugetlb: modular ... |
218 219 |
return ((address - vma->vm_start) >> huge_page_shift(h)) + (vma->vm_pgoff >> huge_page_order(h)); |
e7c4b0bfd huge page private... |
220 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
221 222 223 224 225 |
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... |
226 |
/* |
08fba6998 mm: report the pa... |
227 228 229 230 231 232 233 234 235 236 237 238 239 240 |
* Return the size of the pages allocated when backing a VMA. In the majority * cases this will be same size as used by the page table entries. */ unsigned long vma_kernel_pagesize(struct vm_area_struct *vma) { struct hstate *hstate; if (!is_vm_hugetlb_page(vma)) return PAGE_SIZE; hstate = hstate_vma(vma); return 1UL << (hstate->order + PAGE_SHIFT); } |
f340ca0f0 hugetlbfs: export... |
241 |
EXPORT_SYMBOL_GPL(vma_kernel_pagesize); |
08fba6998 mm: report the pa... |
242 243 |
/* |
3340289dd mm: report the MM... |
244 245 246 247 248 249 250 251 252 253 254 255 256 |
* 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... |
257 258 259 260 261 262 |
* 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... |
263 |
#define HPAGE_RESV_MASK (HPAGE_RESV_OWNER | HPAGE_RESV_UNMAPPED) |
84afd99b8 hugetlb reservati... |
264 |
|
a1e78772d hugetlb: reserve ... |
265 266 267 268 269 270 271 272 273 |
/* * 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... |
274 275 276 277 278 279 280 281 282 |
* * 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 ... |
283 |
*/ |
e7c4b0bfd huge page private... |
284 285 286 287 288 289 290 291 292 293 |
static unsigned long get_vma_private_data(struct vm_area_struct *vma) { return (unsigned long)vma->vm_private_data; } static void set_vma_private_data(struct vm_area_struct *vma, unsigned long value) { vma->vm_private_data = (void *)value; } |
84afd99b8 hugetlb reservati... |
294 295 296 297 |
struct resv_map { struct kref refs; struct list_head regions; }; |
2a4b3ded5 mm: hugetlb.c mak... |
298 |
static struct resv_map *resv_map_alloc(void) |
84afd99b8 hugetlb reservati... |
299 300 301 302 303 304 305 306 307 308 |
{ struct resv_map *resv_map = kmalloc(sizeof(*resv_map), GFP_KERNEL); if (!resv_map) return NULL; kref_init(&resv_map->refs); INIT_LIST_HEAD(&resv_map->regions); return resv_map; } |
2a4b3ded5 mm: hugetlb.c mak... |
309 |
static void resv_map_release(struct kref *ref) |
84afd99b8 hugetlb reservati... |
310 311 312 313 314 315 316 317 318 |
{ struct resv_map *resv_map = container_of(ref, struct resv_map, refs); /* Clear out any active regions before we release the map. */ region_truncate(&resv_map->regions, 0); kfree(resv_map); } static struct resv_map *vma_resv_map(struct vm_area_struct *vma) |
a1e78772d hugetlb: reserve ... |
319 320 |
{ VM_BUG_ON(!is_vm_hugetlb_page(vma)); |
f83a275db mm: account for M... |
321 |
if (!(vma->vm_flags & VM_MAYSHARE)) |
84afd99b8 hugetlb reservati... |
322 323 |
return (struct resv_map *)(get_vma_private_data(vma) & ~HPAGE_RESV_MASK); |
2a4b3ded5 mm: hugetlb.c mak... |
324 |
return NULL; |
a1e78772d hugetlb: reserve ... |
325 |
} |
84afd99b8 hugetlb reservati... |
326 |
static void set_vma_resv_map(struct vm_area_struct *vma, struct resv_map *map) |
a1e78772d hugetlb: reserve ... |
327 328 |
{ VM_BUG_ON(!is_vm_hugetlb_page(vma)); |
f83a275db mm: account for M... |
329 |
VM_BUG_ON(vma->vm_flags & VM_MAYSHARE); |
a1e78772d hugetlb: reserve ... |
330 |
|
84afd99b8 hugetlb reservati... |
331 332 |
set_vma_private_data(vma, (get_vma_private_data(vma) & HPAGE_RESV_MASK) | (unsigned long)map); |
04f2cbe35 hugetlb: guarante... |
333 334 335 336 |
} static void set_vma_resv_flags(struct vm_area_struct *vma, unsigned long flags) { |
04f2cbe35 hugetlb: guarante... |
337 |
VM_BUG_ON(!is_vm_hugetlb_page(vma)); |
f83a275db mm: account for M... |
338 |
VM_BUG_ON(vma->vm_flags & VM_MAYSHARE); |
e7c4b0bfd huge page private... |
339 340 |
set_vma_private_data(vma, get_vma_private_data(vma) | flags); |
04f2cbe35 hugetlb: guarante... |
341 342 343 344 345 |
} static int is_vma_resv_set(struct vm_area_struct *vma, unsigned long flag) { VM_BUG_ON(!is_vm_hugetlb_page(vma)); |
e7c4b0bfd huge page private... |
346 347 |
return (get_vma_private_data(vma) & flag) != 0; |
a1e78772d hugetlb: reserve ... |
348 349 350 |
} /* Decrement the reserved pages in the hugepage pool by one */ |
a55164389 hugetlb: modular ... |
351 352 |
static void decrement_hugepage_resv_vma(struct hstate *h, struct vm_area_struct *vma) |
a1e78772d hugetlb: reserve ... |
353 |
{ |
c37f9fb11 hugetlb: allow hu... |
354 355 |
if (vma->vm_flags & VM_NORESERVE) return; |
f83a275db mm: account for M... |
356 |
if (vma->vm_flags & VM_MAYSHARE) { |
a1e78772d hugetlb: reserve ... |
357 |
/* Shared mappings always use reserves */ |
a55164389 hugetlb: modular ... |
358 |
h->resv_huge_pages--; |
84afd99b8 hugetlb reservati... |
359 |
} else if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) { |
a1e78772d hugetlb: reserve ... |
360 361 362 363 |
/* * Only the process that called mmap() has reserves for * private mappings. */ |
a55164389 hugetlb: modular ... |
364 |
h->resv_huge_pages--; |
a1e78772d hugetlb: reserve ... |
365 366 |
} } |
04f2cbe35 hugetlb: guarante... |
367 |
/* Reset counters to 0 and clear all HPAGE_RESV_* flags */ |
a1e78772d hugetlb: reserve ... |
368 369 370 |
void reset_vma_resv_huge_pages(struct vm_area_struct *vma) { VM_BUG_ON(!is_vm_hugetlb_page(vma)); |
f83a275db mm: account for M... |
371 |
if (!(vma->vm_flags & VM_MAYSHARE)) |
a1e78772d hugetlb: reserve ... |
372 373 374 375 |
vma->vm_private_data = (void *)0; } /* Returns true if the VMA has associated reserve pages */ |
7f09ca51e hugetlb: fix a hu... |
376 |
static int vma_has_reserves(struct vm_area_struct *vma) |
a1e78772d hugetlb: reserve ... |
377 |
{ |
f83a275db mm: account for M... |
378 |
if (vma->vm_flags & VM_MAYSHARE) |
7f09ca51e hugetlb: fix a hu... |
379 380 381 382 |
return 1; if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) return 1; return 0; |
a1e78772d hugetlb: reserve ... |
383 |
} |
69d177c2f hugetlbfs: handle... |
384 385 386 387 388 389 390 391 392 393 394 395 |
static void clear_gigantic_page(struct page *page, unsigned long addr, unsigned long sz) { int i; struct page *p = page; might_sleep(); for (i = 0; i < sz/PAGE_SIZE; i++, p = mem_map_next(p, page, i)) { cond_resched(); clear_user_highpage(p, addr + i * PAGE_SIZE); } } |
a55164389 hugetlb: modular ... |
396 397 |
static void clear_huge_page(struct page *page, unsigned long addr, unsigned long sz) |
79ac6ba40 [PATCH] hugepage:... |
398 399 |
{ int i; |
74dbdd239 mm: hugetlb: fix ... |
400 |
if (unlikely(sz/PAGE_SIZE > MAX_ORDER_NR_PAGES)) { |
ebdd4aea8 hugetlb: fix spar... |
401 402 403 |
clear_gigantic_page(page, addr, sz); return; } |
69d177c2f hugetlbfs: handle... |
404 |
|
79ac6ba40 [PATCH] hugepage:... |
405 |
might_sleep(); |
a55164389 hugetlb: modular ... |
406 |
for (i = 0; i < sz/PAGE_SIZE; i++) { |
79ac6ba40 [PATCH] hugepage:... |
407 |
cond_resched(); |
281e0e3b3 hugetlb: fix clea... |
408 |
clear_user_highpage(page + i, addr + i * PAGE_SIZE); |
79ac6ba40 [PATCH] hugepage:... |
409 410 |
} } |
0ebabb416 hugetlb: redefine... |
411 |
static void copy_user_gigantic_page(struct page *dst, struct page *src, |
69d177c2f hugetlbfs: handle... |
412 413 414 415 416 417 |
unsigned long addr, struct vm_area_struct *vma) { int i; struct hstate *h = hstate_vma(vma); struct page *dst_base = dst; struct page *src_base = src; |
0ebabb416 hugetlb: redefine... |
418 |
|
69d177c2f hugetlbfs: handle... |
419 420 421 422 423 424 425 426 427 |
for (i = 0; i < pages_per_huge_page(h); ) { cond_resched(); copy_user_highpage(dst, src, addr + i*PAGE_SIZE, vma); i++; dst = mem_map_next(dst, dst_base, i); src = mem_map_next(src, src_base, i); } } |
0ebabb416 hugetlb: redefine... |
428 429 |
static void copy_user_huge_page(struct page *dst, struct page *src, |
9de455b20 [PATCH] Pass vma ... |
430 |
unsigned long addr, struct vm_area_struct *vma) |
79ac6ba40 [PATCH] hugepage:... |
431 432 |
{ int i; |
a55164389 hugetlb: modular ... |
433 |
struct hstate *h = hstate_vma(vma); |
79ac6ba40 [PATCH] hugepage:... |
434 |
|
ebdd4aea8 hugetlb: fix spar... |
435 |
if (unlikely(pages_per_huge_page(h) > MAX_ORDER_NR_PAGES)) { |
0ebabb416 hugetlb: redefine... |
436 |
copy_user_gigantic_page(dst, src, addr, vma); |
ebdd4aea8 hugetlb: fix spar... |
437 438 |
return; } |
69d177c2f hugetlbfs: handle... |
439 |
|
79ac6ba40 [PATCH] hugepage:... |
440 |
might_sleep(); |
a55164389 hugetlb: modular ... |
441 |
for (i = 0; i < pages_per_huge_page(h); i++) { |
79ac6ba40 [PATCH] hugepage:... |
442 |
cond_resched(); |
9de455b20 [PATCH] Pass vma ... |
443 |
copy_user_highpage(dst + i, src + i, addr + i*PAGE_SIZE, vma); |
79ac6ba40 [PATCH] hugepage:... |
444 445 |
} } |
0ebabb416 hugetlb: redefine... |
446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 |
static void copy_gigantic_page(struct page *dst, struct page *src) { int i; struct hstate *h = page_hstate(src); struct page *dst_base = dst; struct page *src_base = src; for (i = 0; i < pages_per_huge_page(h); ) { cond_resched(); copy_highpage(dst, src); i++; dst = mem_map_next(dst, dst_base, i); src = mem_map_next(src, src_base, i); } } void copy_huge_page(struct page *dst, struct page *src) { int i; struct hstate *h = page_hstate(src); if (unlikely(pages_per_huge_page(h) > MAX_ORDER_NR_PAGES)) { copy_gigantic_page(dst, src); return; } might_sleep(); for (i = 0; i < pages_per_huge_page(h); i++) { cond_resched(); copy_highpage(dst + i, src + i); } } |
a55164389 hugetlb: modular ... |
479 |
static void enqueue_huge_page(struct hstate *h, struct page *page) |
1da177e4c Linux-2.6.12-rc2 |
480 481 |
{ int nid = page_to_nid(page); |
a55164389 hugetlb: modular ... |
482 483 484 |
list_add(&page->lru, &h->hugepage_freelists[nid]); h->free_huge_pages++; h->free_huge_pages_node[nid]++; |
1da177e4c Linux-2.6.12-rc2 |
485 |
} |
bf50bab2b hugetlb: add allo... |
486 487 488 489 490 491 492 493 |
static struct page *dequeue_huge_page_node(struct hstate *h, int nid) { struct page *page; if (list_empty(&h->hugepage_freelists[nid])) return NULL; page = list_entry(h->hugepage_freelists[nid].next, struct page, lru); list_del(&page->lru); |
a9869b837 hugetlb: move ref... |
494 |
set_page_refcounted(page); |
bf50bab2b hugetlb: add allo... |
495 496 497 498 |
h->free_huge_pages--; h->free_huge_pages_node[nid]--; return page; } |
a55164389 hugetlb: modular ... |
499 500 |
static struct page *dequeue_huge_page_vma(struct hstate *h, struct vm_area_struct *vma, |
04f2cbe35 hugetlb: guarante... |
501 |
unsigned long address, int avoid_reserve) |
1da177e4c Linux-2.6.12-rc2 |
502 |
{ |
1da177e4c Linux-2.6.12-rc2 |
503 |
struct page *page = NULL; |
480eccf9a Fix NUMA Memory P... |
504 |
struct mempolicy *mpol; |
19770b326 mm: filter based ... |
505 |
nodemask_t *nodemask; |
c0ff7453b cpuset,mm: fix no... |
506 |
struct zonelist *zonelist; |
dd1a239f6 mm: have zonelist... |
507 508 |
struct zone *zone; struct zoneref *z; |
1da177e4c Linux-2.6.12-rc2 |
509 |
|
c0ff7453b cpuset,mm: fix no... |
510 511 512 |
get_mems_allowed(); zonelist = huge_zonelist(vma, address, htlb_alloc_mask, &mpol, &nodemask); |
a1e78772d hugetlb: reserve ... |
513 514 515 516 517 |
/* * A child process with MAP_PRIVATE mappings created by their parent * have no page reserves. This check ensures that reservations are * not "stolen". The child may still get SIGKILLed */ |
7f09ca51e hugetlb: fix a hu... |
518 |
if (!vma_has_reserves(vma) && |
a55164389 hugetlb: modular ... |
519 |
h->free_huge_pages - h->resv_huge_pages == 0) |
c0ff7453b cpuset,mm: fix no... |
520 |
goto err; |
a1e78772d hugetlb: reserve ... |
521 |
|
04f2cbe35 hugetlb: guarante... |
522 |
/* If reserves cannot be used, ensure enough pages are in the pool */ |
a55164389 hugetlb: modular ... |
523 |
if (avoid_reserve && h->free_huge_pages - h->resv_huge_pages == 0) |
c0ff7453b cpuset,mm: fix no... |
524 |
goto err;; |
04f2cbe35 hugetlb: guarante... |
525 |
|
19770b326 mm: filter based ... |
526 527 |
for_each_zone_zonelist_nodemask(zone, z, zonelist, MAX_NR_ZONES - 1, nodemask) { |
bf50bab2b hugetlb: add allo... |
528 529 530 531 532 533 534 |
if (cpuset_zone_allowed_softwall(zone, htlb_alloc_mask)) { page = dequeue_huge_page_node(h, zone_to_nid(zone)); if (page) { if (!avoid_reserve) decrement_hugepage_resv_vma(h, vma); break; } |
3abf7afd4 dequeue_huge_page... |
535 |
} |
1da177e4c Linux-2.6.12-rc2 |
536 |
} |
c0ff7453b cpuset,mm: fix no... |
537 |
err: |
52cd3b074 mempolicy: rework... |
538 |
mpol_cond_put(mpol); |
c0ff7453b cpuset,mm: fix no... |
539 |
put_mems_allowed(); |
1da177e4c Linux-2.6.12-rc2 |
540 541 |
return page; } |
a55164389 hugetlb: modular ... |
542 |
static void update_and_free_page(struct hstate *h, struct page *page) |
6af2acb66 hugetlb: Move upd... |
543 544 |
{ int i; |
a55164389 hugetlb: modular ... |
545 |
|
18229df5b hugetlb: pull gig... |
546 |
VM_BUG_ON(h->order >= MAX_ORDER); |
a55164389 hugetlb: modular ... |
547 548 549 |
h->nr_huge_pages--; h->nr_huge_pages_node[page_to_nid(page)]--; for (i = 0; i < pages_per_huge_page(h); i++) { |
6af2acb66 hugetlb: Move upd... |
550 551 552 553 554 555 |
page[i].flags &= ~(1 << PG_locked | 1 << PG_error | 1 << PG_referenced | 1 << PG_dirty | 1 << PG_active | 1 << PG_reserved | 1 << PG_private | 1<< PG_writeback); } set_compound_page_dtor(page, NULL); set_page_refcounted(page); |
7f2e9525b hugetlbfs: common... |
556 |
arch_release_hugepage(page); |
a55164389 hugetlb: modular ... |
557 |
__free_pages(page, huge_page_order(h)); |
6af2acb66 hugetlb: Move upd... |
558 |
} |
e5ff21594 hugetlb: multiple... |
559 560 561 562 563 564 565 566 567 568 |
struct hstate *size_to_hstate(unsigned long size) { struct hstate *h; for_each_hstate(h) { if (huge_page_size(h) == size) return h; } return NULL; } |
27a85ef1b [PATCH] hugepage:... |
569 570 |
static void free_huge_page(struct page *page) { |
a55164389 hugetlb: modular ... |
571 572 573 574 |
/* * Can't pass hstate in here because it is called from the * compound page destructor. */ |
e5ff21594 hugetlb: multiple... |
575 |
struct hstate *h = page_hstate(page); |
7893d1d50 hugetlb: Try to g... |
576 |
int nid = page_to_nid(page); |
c79fb75e5 hugetlb: fix quot... |
577 |
struct address_space *mapping; |
27a85ef1b [PATCH] hugepage:... |
578 |
|
c79fb75e5 hugetlb: fix quot... |
579 |
mapping = (struct address_space *) page_private(page); |
e5df70ab1 hugetlb: ensure w... |
580 |
set_page_private(page, 0); |
23be7468e hugetlb: fix infi... |
581 |
page->mapping = NULL; |
7893d1d50 hugetlb: Try to g... |
582 |
BUG_ON(page_count(page)); |
0fe6e20b9 hugetlb, rmap: ad... |
583 |
BUG_ON(page_mapcount(page)); |
27a85ef1b [PATCH] hugepage:... |
584 585 586 |
INIT_LIST_HEAD(&page->lru); spin_lock(&hugetlb_lock); |
aa888a749 hugetlb: support ... |
587 |
if (h->surplus_huge_pages_node[nid] && huge_page_order(h) < MAX_ORDER) { |
a55164389 hugetlb: modular ... |
588 589 590 |
update_and_free_page(h, page); h->surplus_huge_pages--; h->surplus_huge_pages_node[nid]--; |
7893d1d50 hugetlb: Try to g... |
591 |
} else { |
a55164389 hugetlb: modular ... |
592 |
enqueue_huge_page(h, page); |
7893d1d50 hugetlb: Try to g... |
593 |
} |
27a85ef1b [PATCH] hugepage:... |
594 |
spin_unlock(&hugetlb_lock); |
c79fb75e5 hugetlb: fix quot... |
595 |
if (mapping) |
9a119c056 hugetlb: allow bu... |
596 |
hugetlb_put_quota(mapping, 1); |
27a85ef1b [PATCH] hugepage:... |
597 |
} |
a55164389 hugetlb: modular ... |
598 |
static void prep_new_huge_page(struct hstate *h, struct page *page, int nid) |
b7ba30c67 hugetlb: factor o... |
599 600 601 |
{ set_compound_page_dtor(page, free_huge_page); spin_lock(&hugetlb_lock); |
a55164389 hugetlb: modular ... |
602 603 |
h->nr_huge_pages++; h->nr_huge_pages_node[nid]++; |
b7ba30c67 hugetlb: factor o... |
604 605 606 |
spin_unlock(&hugetlb_lock); put_page(page); /* free it into the hugepage allocator */ } |
20a0307c0 mm: introduce Pag... |
607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 |
static void prep_compound_gigantic_page(struct page *page, unsigned long order) { int i; int nr_pages = 1 << order; struct page *p = page + 1; /* we rely on prep_new_huge_page to set the destructor */ set_compound_order(page, order); __SetPageHead(page); for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) { __SetPageTail(p); p->first_page = page; } } int PageHuge(struct page *page) { compound_page_dtor *dtor; if (!PageCompound(page)) return 0; page = compound_head(page); dtor = get_compound_page_dtor(page); return dtor == free_huge_page; } |
43131e141 HWPOISON, hugetlb... |
634 |
EXPORT_SYMBOL_GPL(PageHuge); |
a55164389 hugetlb: modular ... |
635 |
static struct page *alloc_fresh_huge_page_node(struct hstate *h, int nid) |
1da177e4c Linux-2.6.12-rc2 |
636 |
{ |
1da177e4c Linux-2.6.12-rc2 |
637 |
struct page *page; |
f96efd585 hugetlb: fix race... |
638 |
|
aa888a749 hugetlb: support ... |
639 640 |
if (h->order >= MAX_ORDER) return NULL; |
6484eb3e2 page allocator: d... |
641 |
page = alloc_pages_exact_node(nid, |
551883ae8 page allocator: e... |
642 643 |
htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE| __GFP_REPEAT|__GFP_NOWARN, |
a55164389 hugetlb: modular ... |
644 |
huge_page_order(h)); |
1da177e4c Linux-2.6.12-rc2 |
645 |
if (page) { |
7f2e9525b hugetlbfs: common... |
646 |
if (arch_prepare_hugepage(page)) { |
caff3a2c3 hugetlb: call arc... |
647 |
__free_pages(page, huge_page_order(h)); |
7b8ee84d8 mm: fix integer a... |
648 |
return NULL; |
7f2e9525b hugetlbfs: common... |
649 |
} |
a55164389 hugetlb: modular ... |
650 |
prep_new_huge_page(h, page, nid); |
1da177e4c Linux-2.6.12-rc2 |
651 |
} |
63b4613c3 hugetlb: fix huge... |
652 653 654 |
return page; } |
5ced66c90 hugetlb: abstract... |
655 |
/* |
6ae11b278 hugetlb: add node... |
656 657 658 659 660 |
* common helper functions for hstate_next_node_to_{alloc|free}. * We may have allocated or freed a huge page based on a different * nodes_allowed previously, so h->next_node_to_{alloc|free} might * be outside of *nodes_allowed. Ensure that we use an allowed * node for alloc or free. |
9a76db099 hugetlb: rework h... |
661 |
*/ |
6ae11b278 hugetlb: add node... |
662 |
static int next_node_allowed(int nid, nodemask_t *nodes_allowed) |
9a76db099 hugetlb: rework h... |
663 |
{ |
6ae11b278 hugetlb: add node... |
664 |
nid = next_node(nid, *nodes_allowed); |
9a76db099 hugetlb: rework h... |
665 |
if (nid == MAX_NUMNODES) |
6ae11b278 hugetlb: add node... |
666 |
nid = first_node(*nodes_allowed); |
9a76db099 hugetlb: rework h... |
667 668 669 670 |
VM_BUG_ON(nid >= MAX_NUMNODES); return nid; } |
6ae11b278 hugetlb: add node... |
671 672 673 674 675 676 |
static int get_valid_node_allowed(int nid, nodemask_t *nodes_allowed) { if (!node_isset(nid, *nodes_allowed)) nid = next_node_allowed(nid, nodes_allowed); return nid; } |
9a76db099 hugetlb: rework h... |
677 |
/* |
6ae11b278 hugetlb: add node... |
678 679 680 681 |
* returns the previously saved node ["this node"] from which to * allocate a persistent huge page for the pool and advance the * next node from which to allocate, handling wrap at end of node * mask. |
5ced66c90 hugetlb: abstract... |
682 |
*/ |
6ae11b278 hugetlb: add node... |
683 684 |
static int hstate_next_node_to_alloc(struct hstate *h, nodemask_t *nodes_allowed) |
5ced66c90 hugetlb: abstract... |
685 |
{ |
6ae11b278 hugetlb: add node... |
686 687 688 689 690 691 |
int nid; VM_BUG_ON(!nodes_allowed); nid = get_valid_node_allowed(h->next_nid_to_alloc, nodes_allowed); h->next_nid_to_alloc = next_node_allowed(nid, nodes_allowed); |
9a76db099 hugetlb: rework h... |
692 |
|
9a76db099 hugetlb: rework h... |
693 |
return nid; |
5ced66c90 hugetlb: abstract... |
694 |
} |
6ae11b278 hugetlb: add node... |
695 |
static int alloc_fresh_huge_page(struct hstate *h, nodemask_t *nodes_allowed) |
63b4613c3 hugetlb: fix huge... |
696 697 698 699 700 |
{ struct page *page; int start_nid; int next_nid; int ret = 0; |
6ae11b278 hugetlb: add node... |
701 |
start_nid = hstate_next_node_to_alloc(h, nodes_allowed); |
e8c5c8249 hugetlb: balance ... |
702 |
next_nid = start_nid; |
63b4613c3 hugetlb: fix huge... |
703 704 |
do { |
e8c5c8249 hugetlb: balance ... |
705 |
page = alloc_fresh_huge_page_node(h, next_nid); |
9a76db099 hugetlb: rework h... |
706 |
if (page) { |
63b4613c3 hugetlb: fix huge... |
707 |
ret = 1; |
9a76db099 hugetlb: rework h... |
708 709 |
break; } |
6ae11b278 hugetlb: add node... |
710 |
next_nid = hstate_next_node_to_alloc(h, nodes_allowed); |
9a76db099 hugetlb: rework h... |
711 |
} while (next_nid != start_nid); |
63b4613c3 hugetlb: fix huge... |
712 |
|
3b1163006 Subject: [PATCH] ... |
713 714 715 716 |
if (ret) count_vm_event(HTLB_BUDDY_PGALLOC); else count_vm_event(HTLB_BUDDY_PGALLOC_FAIL); |
63b4613c3 hugetlb: fix huge... |
717 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
718 |
} |
e8c5c8249 hugetlb: balance ... |
719 |
/* |
6ae11b278 hugetlb: add node... |
720 721 722 723 |
* helper for free_pool_huge_page() - return the previously saved * node ["this node"] from which to free a huge page. Advance the * next node id whether or not we find a free huge page to free so * that the next attempt to free addresses the next node. |
e8c5c8249 hugetlb: balance ... |
724 |
*/ |
6ae11b278 hugetlb: add node... |
725 |
static int hstate_next_node_to_free(struct hstate *h, nodemask_t *nodes_allowed) |
e8c5c8249 hugetlb: balance ... |
726 |
{ |
6ae11b278 hugetlb: add node... |
727 728 729 730 731 732 |
int nid; VM_BUG_ON(!nodes_allowed); nid = get_valid_node_allowed(h->next_nid_to_free, nodes_allowed); h->next_nid_to_free = next_node_allowed(nid, nodes_allowed); |
9a76db099 hugetlb: rework h... |
733 |
|
9a76db099 hugetlb: rework h... |
734 |
return nid; |
e8c5c8249 hugetlb: balance ... |
735 736 737 738 739 740 741 742 |
} /* * 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... |
743 744 |
static int free_pool_huge_page(struct hstate *h, nodemask_t *nodes_allowed, bool acct_surplus) |
e8c5c8249 hugetlb: balance ... |
745 746 747 748 |
{ int start_nid; int next_nid; int ret = 0; |
6ae11b278 hugetlb: add node... |
749 |
start_nid = hstate_next_node_to_free(h, nodes_allowed); |
e8c5c8249 hugetlb: balance ... |
750 751 752 |
next_nid = start_nid; do { |
685f34570 hugetlb: use free... |
753 754 755 756 757 758 |
/* * If we're returning unused surplus pages, only examine * nodes with surplus pages. */ if ((!acct_surplus || h->surplus_huge_pages_node[next_nid]) && !list_empty(&h->hugepage_freelists[next_nid])) { |
e8c5c8249 hugetlb: balance ... |
759 760 761 762 763 764 |
struct page *page = list_entry(h->hugepage_freelists[next_nid].next, struct page, lru); list_del(&page->lru); h->free_huge_pages--; h->free_huge_pages_node[next_nid]--; |
685f34570 hugetlb: use free... |
765 766 767 768 |
if (acct_surplus) { h->surplus_huge_pages--; h->surplus_huge_pages_node[next_nid]--; } |
e8c5c8249 hugetlb: balance ... |
769 770 |
update_and_free_page(h, page); ret = 1; |
9a76db099 hugetlb: rework h... |
771 |
break; |
e8c5c8249 hugetlb: balance ... |
772 |
} |
6ae11b278 hugetlb: add node... |
773 |
next_nid = hstate_next_node_to_free(h, nodes_allowed); |
9a76db099 hugetlb: rework h... |
774 |
} while (next_nid != start_nid); |
e8c5c8249 hugetlb: balance ... |
775 776 777 |
return ret; } |
bf50bab2b hugetlb: add allo... |
778 |
static struct page *alloc_buddy_huge_page(struct hstate *h, int nid) |
7893d1d50 hugetlb: Try to g... |
779 780 |
{ struct page *page; |
bf50bab2b hugetlb: add allo... |
781 |
unsigned int r_nid; |
7893d1d50 hugetlb: Try to g... |
782 |
|
aa888a749 hugetlb: support ... |
783 784 |
if (h->order >= MAX_ORDER) return NULL; |
d1c3fb1f8 hugetlb: introduc... |
785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 |
/* * 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 ... |
809 |
if (h->surplus_huge_pages >= h->nr_overcommit_huge_pages) { |
d1c3fb1f8 hugetlb: introduc... |
810 811 812 |
spin_unlock(&hugetlb_lock); return NULL; } else { |
a55164389 hugetlb: modular ... |
813 814 |
h->nr_huge_pages++; h->surplus_huge_pages++; |
d1c3fb1f8 hugetlb: introduc... |
815 816 |
} spin_unlock(&hugetlb_lock); |
bf50bab2b hugetlb: add allo... |
817 818 819 820 821 822 823 824 |
if (nid == NUMA_NO_NODE) page = alloc_pages(htlb_alloc_mask|__GFP_COMP| __GFP_REPEAT|__GFP_NOWARN, huge_page_order(h)); else page = alloc_pages_exact_node(nid, htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE| __GFP_REPEAT|__GFP_NOWARN, huge_page_order(h)); |
d1c3fb1f8 hugetlb: introduc... |
825 |
|
caff3a2c3 hugetlb: call arc... |
826 827 828 829 |
if (page && arch_prepare_hugepage(page)) { __free_pages(page, huge_page_order(h)); return NULL; } |
d1c3fb1f8 hugetlb: introduc... |
830 |
spin_lock(&hugetlb_lock); |
7893d1d50 hugetlb: Try to g... |
831 |
if (page) { |
bf50bab2b hugetlb: add allo... |
832 |
r_nid = page_to_nid(page); |
7893d1d50 hugetlb: Try to g... |
833 |
set_compound_page_dtor(page, free_huge_page); |
d1c3fb1f8 hugetlb: introduc... |
834 835 836 |
/* * We incremented the global counters already */ |
bf50bab2b hugetlb: add allo... |
837 838 |
h->nr_huge_pages_node[r_nid]++; h->surplus_huge_pages_node[r_nid]++; |
3b1163006 Subject: [PATCH] ... |
839 |
__count_vm_event(HTLB_BUDDY_PGALLOC); |
d1c3fb1f8 hugetlb: introduc... |
840 |
} else { |
a55164389 hugetlb: modular ... |
841 842 |
h->nr_huge_pages--; h->surplus_huge_pages--; |
3b1163006 Subject: [PATCH] ... |
843 |
__count_vm_event(HTLB_BUDDY_PGALLOC_FAIL); |
7893d1d50 hugetlb: Try to g... |
844 |
} |
d1c3fb1f8 hugetlb: introduc... |
845 |
spin_unlock(&hugetlb_lock); |
7893d1d50 hugetlb: Try to g... |
846 847 848 |
return page; } |
e4e574b76 hugetlb: Try to g... |
849 |
/* |
bf50bab2b hugetlb: add allo... |
850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 |
* This allocation function is useful in the context where vma is irrelevant. * E.g. soft-offlining uses this function because it only cares physical * address of error page. */ struct page *alloc_huge_page_node(struct hstate *h, int nid) { struct page *page; spin_lock(&hugetlb_lock); page = dequeue_huge_page_node(h, nid); spin_unlock(&hugetlb_lock); if (!page) page = alloc_buddy_huge_page(h, nid); return page; } /* |
e4e574b76 hugetlb: Try to g... |
869 870 871 |
* Increase the hugetlb pool such that it can accomodate a reservation * of size 'delta'. */ |
a55164389 hugetlb: modular ... |
872 |
static int gather_surplus_pages(struct hstate *h, int delta) |
e4e574b76 hugetlb: Try to g... |
873 874 875 876 877 |
{ struct list_head surplus_list; struct page *page, *tmp; int ret, i; int needed, allocated; |
a55164389 hugetlb: modular ... |
878 |
needed = (h->resv_huge_pages + delta) - h->free_huge_pages; |
ac09b3a15 hugetlb: close a ... |
879 |
if (needed <= 0) { |
a55164389 hugetlb: modular ... |
880 |
h->resv_huge_pages += delta; |
e4e574b76 hugetlb: Try to g... |
881 |
return 0; |
ac09b3a15 hugetlb: close a ... |
882 |
} |
e4e574b76 hugetlb: Try to g... |
883 884 885 886 887 888 889 890 |
allocated = 0; INIT_LIST_HEAD(&surplus_list); ret = -ENOMEM; retry: spin_unlock(&hugetlb_lock); for (i = 0; i < needed; i++) { |
bf50bab2b hugetlb: add allo... |
891 |
page = alloc_buddy_huge_page(h, NUMA_NO_NODE); |
a9869b837 hugetlb: move ref... |
892 |
if (!page) |
e4e574b76 hugetlb: Try to g... |
893 894 895 896 897 |
/* * We were not able to allocate enough pages to * satisfy the entire reservation so we free what * we've allocated so far. */ |
e4e574b76 hugetlb: Try to g... |
898 |
goto free; |
e4e574b76 hugetlb: Try to g... |
899 900 901 902 903 904 905 906 907 908 |
list_add(&page->lru, &surplus_list); } allocated += needed; /* * After retaking hugetlb_lock, we need to recalculate 'needed' * because either resv_huge_pages or free_huge_pages may have changed. */ spin_lock(&hugetlb_lock); |
a55164389 hugetlb: modular ... |
909 910 |
needed = (h->resv_huge_pages + delta) - (h->free_huge_pages + allocated); |
e4e574b76 hugetlb: Try to g... |
911 912 913 914 915 916 917 |
if (needed > 0) goto retry; /* * The surplus_list now contains _at_least_ the number of extra pages * needed to accomodate the reservation. Add the appropriate number * of pages to the hugetlb pool and free the extras back to the buddy |
ac09b3a15 hugetlb: close a ... |
918 919 920 |
* 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... |
921 922 |
*/ needed += allocated; |
a55164389 hugetlb: modular ... |
923 |
h->resv_huge_pages += delta; |
e4e574b76 hugetlb: Try to g... |
924 |
ret = 0; |
a9869b837 hugetlb: move ref... |
925 926 |
spin_unlock(&hugetlb_lock); |
19fc3f0ac hugetlb: decrease... |
927 |
/* Free the needed pages to the hugetlb pool */ |
e4e574b76 hugetlb: Try to g... |
928 |
list_for_each_entry_safe(page, tmp, &surplus_list, lru) { |
19fc3f0ac hugetlb: decrease... |
929 930 |
if ((--needed) < 0) break; |
e4e574b76 hugetlb: Try to g... |
931 |
list_del(&page->lru); |
a9869b837 hugetlb: move ref... |
932 933 934 935 936 937 |
/* * This page is now managed by the hugetlb allocator and has * no users -- drop the buddy allocator's reference. */ put_page_testzero(page); VM_BUG_ON(page_count(page)); |
a55164389 hugetlb: modular ... |
938 |
enqueue_huge_page(h, page); |
19fc3f0ac hugetlb: decrease... |
939 940 941 |
} /* Free unnecessary surplus pages to the buddy allocator */ |
a9869b837 hugetlb: move ref... |
942 |
free: |
19fc3f0ac hugetlb: decrease... |
943 |
if (!list_empty(&surplus_list)) { |
19fc3f0ac hugetlb: decrease... |
944 945 |
list_for_each_entry_safe(page, tmp, &surplus_list, lru) { list_del(&page->lru); |
a9869b837 hugetlb: move ref... |
946 |
put_page(page); |
af767cbdd hugetlb: fix dyna... |
947 |
} |
e4e574b76 hugetlb: Try to g... |
948 |
} |
a9869b837 hugetlb: move ref... |
949 |
spin_lock(&hugetlb_lock); |
e4e574b76 hugetlb: Try to g... |
950 951 952 953 954 955 956 957 |
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... |
958 |
* Called with hugetlb_lock held. |
e4e574b76 hugetlb: Try to g... |
959 |
*/ |
a55164389 hugetlb: modular ... |
960 961 |
static void return_unused_surplus_pages(struct hstate *h, unsigned long unused_resv_pages) |
e4e574b76 hugetlb: Try to g... |
962 |
{ |
e4e574b76 hugetlb: Try to g... |
963 |
unsigned long nr_pages; |
ac09b3a15 hugetlb: close a ... |
964 |
/* Uncommit the reservation */ |
a55164389 hugetlb: modular ... |
965 |
h->resv_huge_pages -= unused_resv_pages; |
ac09b3a15 hugetlb: close a ... |
966 |
|
aa888a749 hugetlb: support ... |
967 968 969 |
/* Cannot return gigantic pages currently */ if (h->order >= MAX_ORDER) return; |
a55164389 hugetlb: modular ... |
970 |
nr_pages = min(unused_resv_pages, h->surplus_huge_pages); |
e4e574b76 hugetlb: Try to g... |
971 |
|
685f34570 hugetlb: use free... |
972 973 |
/* * We want to release as many surplus pages as possible, spread |
9b5e5d0fd hugetlb: use only... |
974 975 976 977 978 |
* 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... |
979 980 |
*/ while (nr_pages--) { |
9b5e5d0fd hugetlb: use only... |
981 |
if (!free_pool_huge_page(h, &node_states[N_HIGH_MEMORY], 1)) |
685f34570 hugetlb: use free... |
982 |
break; |
e4e574b76 hugetlb: Try to g... |
983 984 |
} } |
c37f9fb11 hugetlb: allow hu... |
985 986 987 988 989 990 991 992 993 |
/* * Determine if the huge page at addr within the vma has an associated * reservation. Where it does not we will need to logically increase * reservation and actually increase quota before an allocation can occur. * Where any new reservation would be required the reservation change is * prepared, but not committed. Once the page has been quota'd allocated * an instantiated the change should be committed via vma_commit_reservation. * No action is required on failure. */ |
e2f17d945 hugetlb: chg cann... |
994 |
static long vma_needs_reservation(struct hstate *h, |
a55164389 hugetlb: modular ... |
995 |
struct vm_area_struct *vma, unsigned long addr) |
c37f9fb11 hugetlb: allow hu... |
996 997 998 |
{ struct address_space *mapping = vma->vm_file->f_mapping; struct inode *inode = mapping->host; |
f83a275db mm: account for M... |
999 |
if (vma->vm_flags & VM_MAYSHARE) { |
a55164389 hugetlb: modular ... |
1000 |
pgoff_t idx = vma_hugecache_offset(h, vma, addr); |
c37f9fb11 hugetlb: allow hu... |
1001 1002 |
return region_chg(&inode->i_mapping->private_list, idx, idx + 1); |
84afd99b8 hugetlb reservati... |
1003 1004 |
} else if (!is_vma_resv_set(vma, HPAGE_RESV_OWNER)) { return 1; |
c37f9fb11 hugetlb: allow hu... |
1005 |
|
84afd99b8 hugetlb reservati... |
1006 |
} else { |
e2f17d945 hugetlb: chg cann... |
1007 |
long err; |
a55164389 hugetlb: modular ... |
1008 |
pgoff_t idx = vma_hugecache_offset(h, vma, addr); |
84afd99b8 hugetlb reservati... |
1009 1010 1011 1012 1013 1014 1015 |
struct resv_map *reservations = vma_resv_map(vma); err = region_chg(&reservations->regions, idx, idx + 1); if (err < 0) return err; return 0; } |
c37f9fb11 hugetlb: allow hu... |
1016 |
} |
a55164389 hugetlb: modular ... |
1017 1018 |
static void vma_commit_reservation(struct hstate *h, struct vm_area_struct *vma, unsigned long addr) |
c37f9fb11 hugetlb: allow hu... |
1019 1020 1021 |
{ struct address_space *mapping = vma->vm_file->f_mapping; struct inode *inode = mapping->host; |
f83a275db mm: account for M... |
1022 |
if (vma->vm_flags & VM_MAYSHARE) { |
a55164389 hugetlb: modular ... |
1023 |
pgoff_t idx = vma_hugecache_offset(h, vma, addr); |
c37f9fb11 hugetlb: allow hu... |
1024 |
region_add(&inode->i_mapping->private_list, idx, idx + 1); |
84afd99b8 hugetlb reservati... |
1025 1026 |
} else if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) { |
a55164389 hugetlb: modular ... |
1027 |
pgoff_t idx = vma_hugecache_offset(h, vma, addr); |
84afd99b8 hugetlb reservati... |
1028 1029 1030 1031 |
struct resv_map *reservations = vma_resv_map(vma); /* Mark this page used in the map. */ region_add(&reservations->regions, idx, idx + 1); |
c37f9fb11 hugetlb: allow hu... |
1032 1033 |
} } |
a1e78772d hugetlb: reserve ... |
1034 |
static struct page *alloc_huge_page(struct vm_area_struct *vma, |
04f2cbe35 hugetlb: guarante... |
1035 |
unsigned long addr, int avoid_reserve) |
1da177e4c Linux-2.6.12-rc2 |
1036 |
{ |
a55164389 hugetlb: modular ... |
1037 |
struct hstate *h = hstate_vma(vma); |
348ea204c hugetlb: split al... |
1038 |
struct page *page; |
a1e78772d hugetlb: reserve ... |
1039 1040 |
struct address_space *mapping = vma->vm_file->f_mapping; struct inode *inode = mapping->host; |
e2f17d945 hugetlb: chg cann... |
1041 |
long chg; |
a1e78772d hugetlb: reserve ... |
1042 1043 1044 1045 1046 |
/* * Processes that did not create the mapping will have no reserves and * will not have accounted against quota. Check that the quota can be * made before satisfying the allocation |
c37f9fb11 hugetlb: allow hu... |
1047 1048 |
* MAP_NORESERVE mappings may also need pages and quota allocated * if no reserve mapping overlaps. |
a1e78772d hugetlb: reserve ... |
1049 |
*/ |
a55164389 hugetlb: modular ... |
1050 |
chg = vma_needs_reservation(h, vma, addr); |
c37f9fb11 hugetlb: allow hu... |
1051 1052 1053 |
if (chg < 0) return ERR_PTR(chg); if (chg) |
a1e78772d hugetlb: reserve ... |
1054 1055 |
if (hugetlb_get_quota(inode->i_mapping, chg)) return ERR_PTR(-ENOSPC); |
1da177e4c Linux-2.6.12-rc2 |
1056 1057 |
spin_lock(&hugetlb_lock); |
a55164389 hugetlb: modular ... |
1058 |
page = dequeue_huge_page_vma(h, vma, addr, avoid_reserve); |
1da177e4c Linux-2.6.12-rc2 |
1059 |
spin_unlock(&hugetlb_lock); |
b45b5bd65 [PATCH] hugepage:... |
1060 |
|
68842c9b9 hugetlbfs: fix qu... |
1061 |
if (!page) { |
bf50bab2b hugetlb: add allo... |
1062 |
page = alloc_buddy_huge_page(h, NUMA_NO_NODE); |
68842c9b9 hugetlbfs: fix qu... |
1063 |
if (!page) { |
a1e78772d hugetlb: reserve ... |
1064 |
hugetlb_put_quota(inode->i_mapping, chg); |
4a6018f7f hugetlbfs: kill a... |
1065 |
return ERR_PTR(-VM_FAULT_SIGBUS); |
68842c9b9 hugetlbfs: fix qu... |
1066 1067 |
} } |
348ea204c hugetlb: split al... |
1068 |
|
a1e78772d hugetlb: reserve ... |
1069 |
set_page_private(page, (unsigned long) mapping); |
90d8b7e61 hugetlb: enforce ... |
1070 |
|
a55164389 hugetlb: modular ... |
1071 |
vma_commit_reservation(h, vma, addr); |
c37f9fb11 hugetlb: allow hu... |
1072 |
|
90d8b7e61 hugetlb: enforce ... |
1073 |
return page; |
b45b5bd65 [PATCH] hugepage:... |
1074 |
} |
91f47662d mm: hugetlb: remo... |
1075 |
int __weak alloc_bootmem_huge_page(struct hstate *h) |
aa888a749 hugetlb: support ... |
1076 1077 |
{ struct huge_bootmem_page *m; |
9b5e5d0fd hugetlb: use only... |
1078 |
int nr_nodes = nodes_weight(node_states[N_HIGH_MEMORY]); |
aa888a749 hugetlb: support ... |
1079 1080 1081 1082 1083 |
while (nr_nodes) { void *addr; addr = __alloc_bootmem_node_nopanic( |
6ae11b278 hugetlb: add node... |
1084 |
NODE_DATA(hstate_next_node_to_alloc(h, |
9b5e5d0fd hugetlb: use only... |
1085 |
&node_states[N_HIGH_MEMORY])), |
aa888a749 hugetlb: support ... |
1086 1087 1088 1089 1090 1091 1092 1093 1094 |
huge_page_size(h), huge_page_size(h), 0); if (addr) { /* * Use the beginning of the huge page to store the * huge_bootmem_page struct (until gather_bootmem * puts them into the mem_map). */ m = addr; |
91f47662d mm: hugetlb: remo... |
1095 |
goto found; |
aa888a749 hugetlb: support ... |
1096 |
} |
aa888a749 hugetlb: support ... |
1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 |
nr_nodes--; } return 0; found: BUG_ON((unsigned long)virt_to_phys(m) & (huge_page_size(h) - 1)); /* Put them into a private list first because mem_map is not up yet */ list_add(&m->list, &huge_boot_pages); m->hstate = h; return 1; } |
18229df5b hugetlb: pull gig... |
1108 1109 1110 1111 1112 1113 1114 |
static void prep_compound_huge_page(struct page *page, int order) { if (unlikely(order > (MAX_ORDER - 1))) prep_compound_gigantic_page(page, order); else prep_compound_page(page, order); } |
aa888a749 hugetlb: support ... |
1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 |
/* Put bootmem huge pages into the standard lists after mem_map is up */ static void __init gather_bootmem_prealloc(void) { struct huge_bootmem_page *m; list_for_each_entry(m, &huge_boot_pages, list) { struct page *page = virt_to_page(m); struct hstate *h = m->hstate; __ClearPageReserved(page); WARN_ON(page_count(page) != 1); |
18229df5b hugetlb: pull gig... |
1125 |
prep_compound_huge_page(page, h->order); |
aa888a749 hugetlb: support ... |
1126 1127 1128 |
prep_new_huge_page(h, page, page_to_nid(page)); } } |
8faa8b077 hugetlb: support ... |
1129 |
static void __init hugetlb_hstate_alloc_pages(struct hstate *h) |
1da177e4c Linux-2.6.12-rc2 |
1130 1131 |
{ unsigned long i; |
a55164389 hugetlb: modular ... |
1132 |
|
e5ff21594 hugetlb: multiple... |
1133 |
for (i = 0; i < h->max_huge_pages; ++i) { |
aa888a749 hugetlb: support ... |
1134 1135 1136 |
if (h->order >= MAX_ORDER) { if (!alloc_bootmem_huge_page(h)) break; |
9b5e5d0fd hugetlb: use only... |
1137 1138 |
} else if (!alloc_fresh_huge_page(h, &node_states[N_HIGH_MEMORY])) |
1da177e4c Linux-2.6.12-rc2 |
1139 |
break; |
1da177e4c Linux-2.6.12-rc2 |
1140 |
} |
8faa8b077 hugetlb: support ... |
1141 |
h->max_huge_pages = i; |
e5ff21594 hugetlb: multiple... |
1142 1143 1144 1145 1146 1147 1148 |
} static void __init hugetlb_init_hstates(void) { struct hstate *h; for_each_hstate(h) { |
8faa8b077 hugetlb: support ... |
1149 1150 1151 |
/* oversize hugepages were init'ed in early boot */ if (h->order < MAX_ORDER) hugetlb_hstate_alloc_pages(h); |
e5ff21594 hugetlb: multiple... |
1152 1153 |
} } |
4abd32dba hugetlb: printk c... |
1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 |
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... |
1164 1165 1166 1167 1168 |
static void __init report_hugepages(void) { struct hstate *h; for_each_hstate(h) { |
4abd32dba hugetlb: printk c... |
1169 1170 1171 1172 1173 1174 |
char buf[32]; printk(KERN_INFO "HugeTLB registered %s page size, " "pre-allocated %ld pages ", memfmt(buf, huge_page_size(h)), h->free_huge_pages); |
e5ff21594 hugetlb: multiple... |
1175 1176 |
} } |
1da177e4c Linux-2.6.12-rc2 |
1177 |
#ifdef CONFIG_HIGHMEM |
6ae11b278 hugetlb: add node... |
1178 1179 |
static void try_to_free_low(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c Linux-2.6.12-rc2 |
1180 |
{ |
4415cc8df [PATCH] Hugepages... |
1181 |
int i; |
aa888a749 hugetlb: support ... |
1182 1183 |
if (h->order >= MAX_ORDER) return; |
6ae11b278 hugetlb: add node... |
1184 |
for_each_node_mask(i, *nodes_allowed) { |
1da177e4c Linux-2.6.12-rc2 |
1185 |
struct page *page, *next; |
a55164389 hugetlb: modular ... |
1186 1187 1188 |
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... |
1189 |
return; |
1da177e4c Linux-2.6.12-rc2 |
1190 1191 1192 |
if (PageHighMem(page)) continue; list_del(&page->lru); |
e5ff21594 hugetlb: multiple... |
1193 |
update_and_free_page(h, page); |
a55164389 hugetlb: modular ... |
1194 1195 |
h->free_huge_pages--; h->free_huge_pages_node[page_to_nid(page)]--; |
1da177e4c Linux-2.6.12-rc2 |
1196 1197 1198 1199 |
} } } #else |
6ae11b278 hugetlb: add node... |
1200 1201 |
static inline void try_to_free_low(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c Linux-2.6.12-rc2 |
1202 1203 1204 |
{ } #endif |
20a0307c0 mm: introduce Pag... |
1205 1206 1207 1208 1209 |
/* * 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... |
1210 1211 |
static int adjust_pool_surplus(struct hstate *h, nodemask_t *nodes_allowed, int delta) |
20a0307c0 mm: introduce Pag... |
1212 |
{ |
e8c5c8249 hugetlb: balance ... |
1213 |
int start_nid, next_nid; |
20a0307c0 mm: introduce Pag... |
1214 1215 1216 |
int ret = 0; VM_BUG_ON(delta != -1 && delta != 1); |
20a0307c0 mm: introduce Pag... |
1217 |
|
e8c5c8249 hugetlb: balance ... |
1218 |
if (delta < 0) |
6ae11b278 hugetlb: add node... |
1219 |
start_nid = hstate_next_node_to_alloc(h, nodes_allowed); |
e8c5c8249 hugetlb: balance ... |
1220 |
else |
6ae11b278 hugetlb: add node... |
1221 |
start_nid = hstate_next_node_to_free(h, nodes_allowed); |
e8c5c8249 hugetlb: balance ... |
1222 1223 1224 1225 1226 |
next_nid = start_nid; do { int nid = next_nid; if (delta < 0) { |
e8c5c8249 hugetlb: balance ... |
1227 1228 1229 |
/* * To shrink on this node, there must be a surplus page */ |
9a76db099 hugetlb: rework h... |
1230 |
if (!h->surplus_huge_pages_node[nid]) { |
6ae11b278 hugetlb: add node... |
1231 1232 |
next_nid = hstate_next_node_to_alloc(h, nodes_allowed); |
e8c5c8249 hugetlb: balance ... |
1233 |
continue; |
9a76db099 hugetlb: rework h... |
1234 |
} |
e8c5c8249 hugetlb: balance ... |
1235 1236 |
} if (delta > 0) { |
e8c5c8249 hugetlb: balance ... |
1237 1238 1239 1240 |
/* * Surplus cannot exceed the total number of pages */ if (h->surplus_huge_pages_node[nid] >= |
9a76db099 hugetlb: rework h... |
1241 |
h->nr_huge_pages_node[nid]) { |
6ae11b278 hugetlb: add node... |
1242 1243 |
next_nid = hstate_next_node_to_free(h, nodes_allowed); |
e8c5c8249 hugetlb: balance ... |
1244 |
continue; |
9a76db099 hugetlb: rework h... |
1245 |
} |
e8c5c8249 hugetlb: balance ... |
1246 |
} |
20a0307c0 mm: introduce Pag... |
1247 1248 1249 1250 1251 |
h->surplus_huge_pages += delta; h->surplus_huge_pages_node[nid] += delta; ret = 1; break; |
e8c5c8249 hugetlb: balance ... |
1252 |
} while (next_nid != start_nid); |
20a0307c0 mm: introduce Pag... |
1253 |
|
20a0307c0 mm: introduce Pag... |
1254 1255 |
return ret; } |
a55164389 hugetlb: modular ... |
1256 |
#define persistent_huge_pages(h) (h->nr_huge_pages - h->surplus_huge_pages) |
6ae11b278 hugetlb: add node... |
1257 1258 |
static unsigned long set_max_huge_pages(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c Linux-2.6.12-rc2 |
1259 |
{ |
7893d1d50 hugetlb: Try to g... |
1260 |
unsigned long min_count, ret; |
1da177e4c Linux-2.6.12-rc2 |
1261 |
|
aa888a749 hugetlb: support ... |
1262 1263 |
if (h->order >= MAX_ORDER) return h->max_huge_pages; |
7893d1d50 hugetlb: Try to g... |
1264 1265 1266 1267 |
/* * 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... |
1268 1269 1270 1271 1272 1273 |
* * 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... |
1274 |
*/ |
1da177e4c Linux-2.6.12-rc2 |
1275 |
spin_lock(&hugetlb_lock); |
a55164389 hugetlb: modular ... |
1276 |
while (h->surplus_huge_pages && count > persistent_huge_pages(h)) { |
6ae11b278 hugetlb: add node... |
1277 |
if (!adjust_pool_surplus(h, nodes_allowed, -1)) |
7893d1d50 hugetlb: Try to g... |
1278 1279 |
break; } |
a55164389 hugetlb: modular ... |
1280 |
while (count > persistent_huge_pages(h)) { |
7893d1d50 hugetlb: Try to g... |
1281 1282 1283 1284 1285 1286 |
/* * If this allocation races such that we no longer need the * page, free_huge_page will handle it by freeing the page * and reducing the surplus. */ spin_unlock(&hugetlb_lock); |
6ae11b278 hugetlb: add node... |
1287 |
ret = alloc_fresh_huge_page(h, nodes_allowed); |
7893d1d50 hugetlb: Try to g... |
1288 1289 1290 |
spin_lock(&hugetlb_lock); if (!ret) goto out; |
536240f2b hugetlb: abort a ... |
1291 1292 1293 |
/* Bail for signals. Probably ctrl-c from user */ if (signal_pending(current)) goto out; |
7893d1d50 hugetlb: Try to g... |
1294 |
} |
7893d1d50 hugetlb: Try to g... |
1295 1296 1297 1298 1299 1300 1301 |
/* * 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... |
1302 1303 1304 1305 1306 1307 1308 1309 |
* * 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... |
1310 |
*/ |
a55164389 hugetlb: modular ... |
1311 |
min_count = h->resv_huge_pages + h->nr_huge_pages - h->free_huge_pages; |
6b0c880df hugetlb: fix pool... |
1312 |
min_count = max(count, min_count); |
6ae11b278 hugetlb: add node... |
1313 |
try_to_free_low(h, min_count, nodes_allowed); |
a55164389 hugetlb: modular ... |
1314 |
while (min_count < persistent_huge_pages(h)) { |
6ae11b278 hugetlb: add node... |
1315 |
if (!free_pool_huge_page(h, nodes_allowed, 0)) |
1da177e4c Linux-2.6.12-rc2 |
1316 |
break; |
1da177e4c Linux-2.6.12-rc2 |
1317 |
} |
a55164389 hugetlb: modular ... |
1318 |
while (count < persistent_huge_pages(h)) { |
6ae11b278 hugetlb: add node... |
1319 |
if (!adjust_pool_surplus(h, nodes_allowed, 1)) |
7893d1d50 hugetlb: Try to g... |
1320 1321 1322 |
break; } out: |
a55164389 hugetlb: modular ... |
1323 |
ret = persistent_huge_pages(h); |
1da177e4c Linux-2.6.12-rc2 |
1324 |
spin_unlock(&hugetlb_lock); |
7893d1d50 hugetlb: Try to g... |
1325 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
1326 |
} |
a34378701 hugetlb: new sysf... |
1327 1328 1329 1330 1331 1332 1333 1334 1335 |
#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 ... |
1336 1337 1338 |
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... |
1339 1340 |
{ int i; |
9a3052306 hugetlb: add per ... |
1341 |
|
a34378701 hugetlb: new sysf... |
1342 |
for (i = 0; i < HUGE_MAX_HSTATE; i++) |
9a3052306 hugetlb: add per ... |
1343 1344 1345 |
if (hstate_kobjs[i] == kobj) { if (nidp) *nidp = NUMA_NO_NODE; |
a34378701 hugetlb: new sysf... |
1346 |
return &hstates[i]; |
9a3052306 hugetlb: add per ... |
1347 1348 1349 |
} return kobj_to_node_hstate(kobj, nidp); |
a34378701 hugetlb: new sysf... |
1350 |
} |
06808b082 hugetlb: derive h... |
1351 |
static ssize_t nr_hugepages_show_common(struct kobject *kobj, |
a34378701 hugetlb: new sysf... |
1352 1353 |
struct kobj_attribute *attr, char *buf) { |
9a3052306 hugetlb: add per ... |
1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 |
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... |
1366 |
} |
06808b082 hugetlb: derive h... |
1367 1368 1369 |
static ssize_t nr_hugepages_store_common(bool obey_mempolicy, struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t len) |
a34378701 hugetlb: new sysf... |
1370 1371 |
{ int err; |
9a3052306 hugetlb: add per ... |
1372 |
int nid; |
06808b082 hugetlb: derive h... |
1373 |
unsigned long count; |
9a3052306 hugetlb: add per ... |
1374 |
struct hstate *h; |
bad44b5be mm: add gfp flags... |
1375 |
NODEMASK_ALLOC(nodemask_t, nodes_allowed, GFP_KERNEL | __GFP_NORETRY); |
a34378701 hugetlb: new sysf... |
1376 |
|
06808b082 hugetlb: derive h... |
1377 |
err = strict_strtoul(buf, 10, &count); |
a34378701 hugetlb: new sysf... |
1378 1379 |
if (err) return 0; |
9a3052306 hugetlb: add per ... |
1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 |
h = kobj_to_hstate(kobj, &nid); if (nid == NUMA_NO_NODE) { /* * global hstate attribute */ if (!(obey_mempolicy && init_nodemask_of_mempolicy(nodes_allowed))) { NODEMASK_FREE(nodes_allowed); nodes_allowed = &node_states[N_HIGH_MEMORY]; } } else if (nodes_allowed) { /* * per node hstate attribute: adjust count to global, * but restrict alloc/free to the specified node. */ count += h->nr_huge_pages - h->nr_huge_pages_node[nid]; init_nodemask_of_node(nodes_allowed, nid); } else nodes_allowed = &node_states[N_HIGH_MEMORY]; |
06808b082 hugetlb: derive h... |
1399 |
h->max_huge_pages = set_max_huge_pages(h, count, nodes_allowed); |
a34378701 hugetlb: new sysf... |
1400 |
|
9b5e5d0fd hugetlb: use only... |
1401 |
if (nodes_allowed != &node_states[N_HIGH_MEMORY]) |
06808b082 hugetlb: derive h... |
1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 |
NODEMASK_FREE(nodes_allowed); return len; } static ssize_t nr_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return nr_hugepages_show_common(kobj, attr, buf); } static ssize_t nr_hugepages_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t len) { return nr_hugepages_store_common(false, kobj, attr, buf, len); |
a34378701 hugetlb: new sysf... |
1417 1418 |
} HSTATE_ATTR(nr_hugepages); |
06808b082 hugetlb: derive h... |
1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 |
#ifdef CONFIG_NUMA /* * hstate attribute for optionally mempolicy-based constraint on persistent * huge page alloc/free. */ static ssize_t nr_hugepages_mempolicy_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return nr_hugepages_show_common(kobj, attr, buf); } static ssize_t nr_hugepages_mempolicy_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t len) { return nr_hugepages_store_common(true, kobj, attr, buf, len); } HSTATE_ATTR(nr_hugepages_mempolicy); #endif |
a34378701 hugetlb: new sysf... |
1438 1439 1440 |
static ssize_t nr_overcommit_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
9a3052306 hugetlb: add per ... |
1441 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701 hugetlb: new sysf... |
1442 1443 1444 1445 1446 1447 1448 1449 |
return sprintf(buf, "%lu ", h->nr_overcommit_huge_pages); } 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 ... |
1450 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701 hugetlb: new sysf... |
1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 |
err = strict_strtoul(buf, 10, &input); if (err) return 0; 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 ... |
1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 |
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... |
1479 1480 1481 1482 1483 1484 |
} HSTATE_ATTR_RO(free_hugepages); static ssize_t resv_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
9a3052306 hugetlb: add per ... |
1485 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701 hugetlb: new sysf... |
1486 1487 1488 1489 1490 1491 1492 1493 |
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 ... |
1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 |
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... |
1506 1507 1508 1509 1510 1511 1512 1513 1514 |
} 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... |
1515 1516 1517 |
#ifdef CONFIG_NUMA &nr_hugepages_mempolicy_attr.attr, #endif |
a34378701 hugetlb: new sysf... |
1518 1519 1520 1521 1522 1523 |
NULL, }; static struct attribute_group hstate_attr_group = { .attrs = hstate_attrs, }; |
094e9539b hugetlb: fix sect... |
1524 1525 1526 |
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... |
1527 1528 |
{ int retval; |
9a3052306 hugetlb: add per ... |
1529 |
int hi = h - hstates; |
a34378701 hugetlb: new sysf... |
1530 |
|
9a3052306 hugetlb: add per ... |
1531 1532 |
hstate_kobjs[hi] = kobject_create_and_add(h->name, parent); if (!hstate_kobjs[hi]) |
a34378701 hugetlb: new sysf... |
1533 |
return -ENOMEM; |
9a3052306 hugetlb: add per ... |
1534 |
retval = sysfs_create_group(hstate_kobjs[hi], hstate_attr_group); |
a34378701 hugetlb: new sysf... |
1535 |
if (retval) |
9a3052306 hugetlb: add per ... |
1536 |
kobject_put(hstate_kobjs[hi]); |
a34378701 hugetlb: new sysf... |
1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 |
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 ... |
1551 1552 |
err = hugetlb_sysfs_add_hstate(h, hugepages_kobj, hstate_kobjs, &hstate_attr_group); |
a34378701 hugetlb: new sysf... |
1553 1554 1555 1556 1557 |
if (err) printk(KERN_ERR "Hugetlb: Unable to add hstate %s", h->name); } } |
9a3052306 hugetlb: add per ... |
1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 |
#ifdef CONFIG_NUMA /* * node_hstate/s - associate per node hstate attributes, via their kobjects, * with node sysdevs in node_devices[] using a parallel array. The array * index of a node sysdev or _hstate == node id. * This is here to avoid any static dependency of the node sysdev driver, in * the base kernel, on the hugetlb module. */ struct node_hstate { struct kobject *hugepages_kobj; struct kobject *hstate_kobjs[HUGE_MAX_HSTATE]; }; struct node_hstate node_hstates[MAX_NUMNODES]; /* * A subset of global hstate attributes for node sysdevs */ static struct attribute *per_node_hstate_attrs[] = { &nr_hugepages_attr.attr, &free_hugepages_attr.attr, &surplus_hugepages_attr.attr, NULL, }; static struct attribute_group per_node_hstate_attr_group = { .attrs = per_node_hstate_attrs, }; /* * kobj_to_node_hstate - lookup global hstate for node sysdev hstate attr kobj. * Returns node id via non-NULL nidp. */ static struct hstate *kobj_to_node_hstate(struct kobject *kobj, int *nidp) { int nid; for (nid = 0; nid < nr_node_ids; nid++) { struct node_hstate *nhs = &node_hstates[nid]; int i; for (i = 0; i < HUGE_MAX_HSTATE; i++) if (nhs->hstate_kobjs[i] == kobj) { if (nidp) *nidp = nid; return &hstates[i]; } } BUG(); return NULL; } /* * Unregister hstate attributes from a single node sysdev. * No-op if no hstate attributes attached. */ void hugetlb_unregister_node(struct node *node) { struct hstate *h; struct node_hstate *nhs = &node_hstates[node->sysdev.id]; if (!nhs->hugepages_kobj) |
9b5e5d0fd hugetlb: use only... |
1620 |
return; /* no hstate attributes */ |
9a3052306 hugetlb: add per ... |
1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 |
for_each_hstate(h) if (nhs->hstate_kobjs[h - hstates]) { kobject_put(nhs->hstate_kobjs[h - hstates]); nhs->hstate_kobjs[h - hstates] = NULL; } kobject_put(nhs->hugepages_kobj); nhs->hugepages_kobj = NULL; } /* * hugetlb module exit: unregister hstate attributes from node sysdevs * that have them. */ static void hugetlb_unregister_all_nodes(void) { int nid; /* * disable node sysdev registrations. */ register_hugetlbfs_with_node(NULL, NULL); /* * remove hstate attributes from any nodes that have them. */ for (nid = 0; nid < nr_node_ids; nid++) hugetlb_unregister_node(&node_devices[nid]); } /* * Register hstate attributes for a single node sysdev. * No-op if attributes already registered. */ void hugetlb_register_node(struct node *node) { struct hstate *h; struct node_hstate *nhs = &node_hstates[node->sysdev.id]; int err; if (nhs->hugepages_kobj) return; /* already allocated */ nhs->hugepages_kobj = kobject_create_and_add("hugepages", &node->sysdev.kobj); if (!nhs->hugepages_kobj) return; for_each_hstate(h) { err = hugetlb_sysfs_add_hstate(h, nhs->hugepages_kobj, nhs->hstate_kobjs, &per_node_hstate_attr_group); if (err) { printk(KERN_ERR "Hugetlb: Unable to add hstate %s" " for node %d ", h->name, node->sysdev.id); hugetlb_unregister_node(node); break; } } } /* |
9b5e5d0fd hugetlb: use only... |
1686 1687 1688 |
* hugetlb init time: register hstate attributes for all registered node * sysdevs of nodes that have memory. All on-line nodes should have * registered their associated sysdev by this time. |
9a3052306 hugetlb: add per ... |
1689 1690 1691 1692 |
*/ static void hugetlb_register_all_nodes(void) { int nid; |
9b5e5d0fd hugetlb: use only... |
1693 |
for_each_node_state(nid, N_HIGH_MEMORY) { |
9a3052306 hugetlb: add per ... |
1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 |
struct node *node = &node_devices[nid]; if (node->sysdev.id == nid) hugetlb_register_node(node); } /* * Let the node sysdev driver know we're here so it can * [un]register hstate attributes on node hotplug. */ register_hugetlbfs_with_node(hugetlb_register_node, hugetlb_unregister_node); } #else /* !CONFIG_NUMA */ static struct hstate *kobj_to_node_hstate(struct kobject *kobj, int *nidp) { BUG(); if (nidp) *nidp = -1; return NULL; } static void hugetlb_unregister_all_nodes(void) { } static void hugetlb_register_all_nodes(void) { } #endif |
a34378701 hugetlb: new sysf... |
1721 1722 1723 |
static void __exit hugetlb_exit(void) { struct hstate *h; |
9a3052306 hugetlb: add per ... |
1724 |
hugetlb_unregister_all_nodes(); |
a34378701 hugetlb: new sysf... |
1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 |
for_each_hstate(h) { kobject_put(hstate_kobjs[h - hstates]); } kobject_put(hugepages_kobj); } module_exit(hugetlb_exit); static int __init hugetlb_init(void) { |
0ef89d25d mm/hugetlb: don't... |
1735 1736 1737 1738 1739 1740 |
/* Some platform decide whether they support huge pages at boot * time. On these, such as powerpc, HPAGE_SHIFT is set to 0 when * there is no such support */ if (HPAGE_SHIFT == 0) return 0; |
a34378701 hugetlb: new sysf... |
1741 |
|
e11bfbfcb hugetlb: override... |
1742 1743 1744 1745 |
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... |
1746 |
} |
e11bfbfcb hugetlb: override... |
1747 1748 1749 |
default_hstate_idx = size_to_hstate(default_hstate_size) - hstates; if (default_hstate_max_huge_pages) default_hstate.max_huge_pages = default_hstate_max_huge_pages; |
a34378701 hugetlb: new sysf... |
1750 1751 |
hugetlb_init_hstates(); |
aa888a749 hugetlb: support ... |
1752 |
gather_bootmem_prealloc(); |
a34378701 hugetlb: new sysf... |
1753 1754 1755 |
report_hugepages(); hugetlb_sysfs_init(); |
9a3052306 hugetlb: add per ... |
1756 |
hugetlb_register_all_nodes(); |
a34378701 hugetlb: new sysf... |
1757 1758 1759 1760 1761 1762 1763 1764 |
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 ... |
1765 |
unsigned long i; |
a34378701 hugetlb: new sysf... |
1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 |
if (size_to_hstate(PAGE_SIZE << order)) { printk(KERN_WARNING "hugepagesz= specified twice, ignoring "); return; } BUG_ON(max_hstate >= HUGE_MAX_HSTATE); BUG_ON(order == 0); h = &hstates[max_hstate++]; h->order = order; h->mask = ~((1ULL << (order + PAGE_SHIFT)) - 1); |
8faa8b077 hugetlb: support ... |
1776 1777 1778 1779 |
h->nr_huge_pages = 0; h->free_huge_pages = 0; for (i = 0; i < MAX_NUMNODES; ++i) INIT_LIST_HEAD(&h->hugepage_freelists[i]); |
9b5e5d0fd hugetlb: use only... |
1780 1781 |
h->next_nid_to_alloc = first_node(node_states[N_HIGH_MEMORY]); h->next_nid_to_free = first_node(node_states[N_HIGH_MEMORY]); |
a34378701 hugetlb: new sysf... |
1782 1783 |
snprintf(h->name, HSTATE_NAME_LEN, "hugepages-%lukB", huge_page_size(h)/1024); |
8faa8b077 hugetlb: support ... |
1784 |
|
a34378701 hugetlb: new sysf... |
1785 1786 |
parsed_hstate = h; } |
e11bfbfcb hugetlb: override... |
1787 |
static int __init hugetlb_nrpages_setup(char *s) |
a34378701 hugetlb: new sysf... |
1788 1789 |
{ unsigned long *mhp; |
8faa8b077 hugetlb: support ... |
1790 |
static unsigned long *last_mhp; |
a34378701 hugetlb: new sysf... |
1791 1792 1793 1794 1795 1796 1797 1798 1799 |
/* * !max_hstate means we haven't parsed a hugepagesz= parameter yet, * so this hugepages= parameter goes to the "default hstate". */ if (!max_hstate) mhp = &default_hstate_max_huge_pages; else mhp = &parsed_hstate->max_huge_pages; |
8faa8b077 hugetlb: support ... |
1800 1801 1802 1803 1804 1805 |
if (mhp == last_mhp) { printk(KERN_WARNING "hugepages= specified twice without " "interleaving hugepagesz=, ignoring "); return 1; } |
a34378701 hugetlb: new sysf... |
1806 1807 |
if (sscanf(s, "%lu", mhp) <= 0) *mhp = 0; |
8faa8b077 hugetlb: support ... |
1808 1809 1810 1811 1812 1813 1814 1815 1816 |
/* * Global state is always initialized later in hugetlb_init. * But we need to allocate >= MAX_ORDER hstates here early to still * use the bootmem allocator. */ if (max_hstate && parsed_hstate->order >= MAX_ORDER) hugetlb_hstate_alloc_pages(parsed_hstate); last_mhp = mhp; |
a34378701 hugetlb: new sysf... |
1817 1818 |
return 1; } |
e11bfbfcb hugetlb: override... |
1819 1820 1821 1822 1823 1824 1825 1826 |
__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... |
1827 |
|
8a2134605 hugetlb: fix CONF... |
1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 |
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... |
1840 1841 1842 |
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 |
1843 |
{ |
e5ff21594 hugetlb: multiple... |
1844 1845 1846 1847 1848 1849 1850 1851 |
struct hstate *h = &default_hstate; unsigned long tmp; if (!write) tmp = h->max_huge_pages; table->data = &tmp; table->maxlen = sizeof(unsigned long); |
8d65af789 sysctl: remove "s... |
1852 |
proc_doulongvec_minmax(table, write, buffer, length, ppos); |
e5ff21594 hugetlb: multiple... |
1853 |
|
06808b082 hugetlb: derive h... |
1854 |
if (write) { |
bad44b5be mm: add gfp flags... |
1855 1856 |
NODEMASK_ALLOC(nodemask_t, nodes_allowed, GFP_KERNEL | __GFP_NORETRY); |
06808b082 hugetlb: derive h... |
1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 |
if (!(obey_mempolicy && init_nodemask_of_mempolicy(nodes_allowed))) { NODEMASK_FREE(nodes_allowed); nodes_allowed = &node_states[N_HIGH_MEMORY]; } h->max_huge_pages = set_max_huge_pages(h, tmp, nodes_allowed); if (nodes_allowed != &node_states[N_HIGH_MEMORY]) NODEMASK_FREE(nodes_allowed); } |
e5ff21594 hugetlb: multiple... |
1867 |
|
1da177e4c Linux-2.6.12-rc2 |
1868 1869 |
return 0; } |
396faf030 Allow huge page a... |
1870 |
|
06808b082 hugetlb: derive h... |
1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 |
int hugetlb_sysctl_handler(struct ctl_table *table, int write, void __user *buffer, size_t *length, loff_t *ppos) { return hugetlb_sysctl_handler_common(false, table, write, buffer, length, ppos); } #ifdef CONFIG_NUMA int hugetlb_mempolicy_sysctl_handler(struct ctl_table *table, int write, void __user *buffer, size_t *length, loff_t *ppos) { return hugetlb_sysctl_handler_common(true, table, write, buffer, length, ppos); } #endif /* CONFIG_NUMA */ |
396faf030 Allow huge page a... |
1887 |
int hugetlb_treat_movable_handler(struct ctl_table *table, int write, |
8d65af789 sysctl: remove "s... |
1888 |
void __user *buffer, |
396faf030 Allow huge page a... |
1889 1890 |
size_t *length, loff_t *ppos) { |
8d65af789 sysctl: remove "s... |
1891 |
proc_dointvec(table, write, buffer, length, ppos); |
396faf030 Allow huge page a... |
1892 1893 1894 1895 1896 1897 |
if (hugepages_treat_as_movable) htlb_alloc_mask = GFP_HIGHUSER_MOVABLE; else htlb_alloc_mask = GFP_HIGHUSER; return 0; } |
a3d0c6aa1 hugetlb: add lock... |
1898 |
int hugetlb_overcommit_handler(struct ctl_table *table, int write, |
8d65af789 sysctl: remove "s... |
1899 |
void __user *buffer, |
a3d0c6aa1 hugetlb: add lock... |
1900 1901 |
size_t *length, loff_t *ppos) { |
a55164389 hugetlb: modular ... |
1902 |
struct hstate *h = &default_hstate; |
e5ff21594 hugetlb: multiple... |
1903 1904 1905 1906 1907 1908 1909 |
unsigned long tmp; if (!write) tmp = h->nr_overcommit_huge_pages; table->data = &tmp; table->maxlen = sizeof(unsigned long); |
8d65af789 sysctl: remove "s... |
1910 |
proc_doulongvec_minmax(table, write, buffer, length, ppos); |
e5ff21594 hugetlb: multiple... |
1911 1912 1913 1914 1915 1916 |
if (write) { spin_lock(&hugetlb_lock); h->nr_overcommit_huge_pages = tmp; spin_unlock(&hugetlb_lock); } |
a3d0c6aa1 hugetlb: add lock... |
1917 1918 |
return 0; } |
1da177e4c Linux-2.6.12-rc2 |
1919 |
#endif /* CONFIG_SYSCTL */ |
e1759c215 proc: switch /pro... |
1920 |
void hugetlb_report_meminfo(struct seq_file *m) |
1da177e4c Linux-2.6.12-rc2 |
1921 |
{ |
a55164389 hugetlb: modular ... |
1922 |
struct hstate *h = &default_hstate; |
e1759c215 proc: switch /pro... |
1923 |
seq_printf(m, |
4f98a2fee vmscan: split LRU... |
1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 |
"HugePages_Total: %5lu " "HugePages_Free: %5lu " "HugePages_Rsvd: %5lu " "HugePages_Surp: %5lu " "Hugepagesize: %8lu kB ", |
a55164389 hugetlb: modular ... |
1934 1935 1936 1937 1938 |
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 |
1939 1940 1941 1942 |
} int hugetlb_report_node_meminfo(int nid, char *buf) { |
a55164389 hugetlb: modular ... |
1943 |
struct hstate *h = &default_hstate; |
1da177e4c Linux-2.6.12-rc2 |
1944 1945 1946 |
return sprintf(buf, "Node %d HugePages_Total: %5u " |
a1de09195 hugetlb: indicate... |
1947 1948 1949 1950 |
"Node %d HugePages_Free: %5u " "Node %d HugePages_Surp: %5u ", |
a55164389 hugetlb: modular ... |
1951 1952 1953 |
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 |
1954 |
} |
1da177e4c Linux-2.6.12-rc2 |
1955 1956 1957 |
/* Return the number pages of memory we physically have, in PAGE_SIZE units. */ unsigned long hugetlb_total_pages(void) { |
a55164389 hugetlb: modular ... |
1958 1959 |
struct hstate *h = &default_hstate; return h->nr_huge_pages * pages_per_huge_page(h); |
1da177e4c Linux-2.6.12-rc2 |
1960 |
} |
1da177e4c Linux-2.6.12-rc2 |
1961 |
|
a55164389 hugetlb: modular ... |
1962 |
static int hugetlb_acct_memory(struct hstate *h, long delta) |
fc1b8a73d hugetlb: move hug... |
1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 |
{ 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 ... |
1985 |
if (gather_surplus_pages(h, delta) < 0) |
fc1b8a73d hugetlb: move hug... |
1986 |
goto out; |
a55164389 hugetlb: modular ... |
1987 1988 |
if (delta > cpuset_mems_nr(h->free_huge_pages_node)) { return_unused_surplus_pages(h, delta); |
fc1b8a73d hugetlb: move hug... |
1989 1990 1991 1992 1993 1994 |
goto out; } } ret = 0; if (delta < 0) |
a55164389 hugetlb: modular ... |
1995 |
return_unused_surplus_pages(h, (unsigned long) -delta); |
fc1b8a73d hugetlb: move hug... |
1996 1997 1998 1999 2000 |
out: spin_unlock(&hugetlb_lock); return ret; } |
84afd99b8 hugetlb reservati... |
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 |
static void hugetlb_vm_op_open(struct vm_area_struct *vma) { struct resv_map *reservations = vma_resv_map(vma); /* * This new VMA should share its siblings reservation map if present. * The VMA will only ever have a valid reservation map pointer where * it is being copied for another still existing VMA. As that VMA * has a reference to the reservation map it cannot dissappear until * after this open call completes. It is therefore safe to take a * new reference here without additional locking. */ if (reservations) kref_get(&reservations->refs); } |
a1e78772d hugetlb: reserve ... |
2016 2017 |
static void hugetlb_vm_op_close(struct vm_area_struct *vma) { |
a55164389 hugetlb: modular ... |
2018 |
struct hstate *h = hstate_vma(vma); |
84afd99b8 hugetlb reservati... |
2019 2020 2021 2022 2023 2024 |
struct resv_map *reservations = vma_resv_map(vma); unsigned long reserve; unsigned long start; unsigned long end; if (reservations) { |
a55164389 hugetlb: modular ... |
2025 2026 |
start = vma_hugecache_offset(h, vma, vma->vm_start); end = vma_hugecache_offset(h, vma, vma->vm_end); |
84afd99b8 hugetlb reservati... |
2027 2028 2029 2030 2031 |
reserve = (end - start) - region_count(&reservations->regions, start, end); kref_put(&reservations->refs, resv_map_release); |
7251ff78b hugetlb: quota is... |
2032 |
if (reserve) { |
a55164389 hugetlb: modular ... |
2033 |
hugetlb_acct_memory(h, -reserve); |
7251ff78b hugetlb: quota is... |
2034 2035 |
hugetlb_put_quota(vma->vm_file->f_mapping, reserve); } |
84afd99b8 hugetlb reservati... |
2036 |
} |
a1e78772d hugetlb: reserve ... |
2037 |
} |
1da177e4c Linux-2.6.12-rc2 |
2038 2039 2040 2041 2042 2043 |
/* * 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... |
2044 |
static int hugetlb_vm_op_fault(struct vm_area_struct *vma, struct vm_fault *vmf) |
1da177e4c Linux-2.6.12-rc2 |
2045 2046 |
{ BUG(); |
d0217ac04 mm: fault feedbac... |
2047 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
2048 |
} |
f0f37e2f7 const: mark struc... |
2049 |
const struct vm_operations_struct hugetlb_vm_ops = { |
d0217ac04 mm: fault feedbac... |
2050 |
.fault = hugetlb_vm_op_fault, |
84afd99b8 hugetlb reservati... |
2051 |
.open = hugetlb_vm_op_open, |
a1e78772d hugetlb: reserve ... |
2052 |
.close = hugetlb_vm_op_close, |
1da177e4c Linux-2.6.12-rc2 |
2053 |
}; |
1e8f889b1 [PATCH] Hugetlb: ... |
2054 2055 |
static pte_t make_huge_pte(struct vm_area_struct *vma, struct page *page, int writable) |
63551ae0f [PATCH] Hugepage ... |
2056 2057 |
{ pte_t entry; |
1e8f889b1 [PATCH] Hugetlb: ... |
2058 |
if (writable) { |
63551ae0f [PATCH] Hugepage ... |
2059 2060 2061 |
entry = pte_mkwrite(pte_mkdirty(mk_pte(page, vma->vm_page_prot))); } else { |
7f2e9525b hugetlbfs: common... |
2062 |
entry = huge_pte_wrprotect(mk_pte(page, vma->vm_page_prot)); |
63551ae0f [PATCH] Hugepage ... |
2063 2064 2065 2066 2067 2068 |
} entry = pte_mkyoung(entry); entry = pte_mkhuge(entry); return entry; } |
1e8f889b1 [PATCH] Hugetlb: ... |
2069 2070 2071 2072 |
static void set_huge_ptep_writable(struct vm_area_struct *vma, unsigned long address, pte_t *ptep) { pte_t entry; |
7f2e9525b hugetlbfs: common... |
2073 2074 |
entry = pte_mkwrite(pte_mkdirty(huge_ptep_get(ptep))); if (huge_ptep_set_access_flags(vma, address, ptep, entry, 1)) { |
4b3073e1c MM: Pass a PTE po... |
2075 |
update_mmu_cache(vma, address, ptep); |
8dab5241d Rework ptep_set_a... |
2076 |
} |
1e8f889b1 [PATCH] Hugetlb: ... |
2077 |
} |
63551ae0f [PATCH] Hugepage ... |
2078 2079 2080 2081 2082 |
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... |
2083 |
unsigned long addr; |
1e8f889b1 [PATCH] Hugetlb: ... |
2084 |
int cow; |
a55164389 hugetlb: modular ... |
2085 2086 |
struct hstate *h = hstate_vma(vma); unsigned long sz = huge_page_size(h); |
1e8f889b1 [PATCH] Hugetlb: ... |
2087 2088 |
cow = (vma->vm_flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE; |
63551ae0f [PATCH] Hugepage ... |
2089 |
|
a55164389 hugetlb: modular ... |
2090 |
for (addr = vma->vm_start; addr < vma->vm_end; addr += sz) { |
c74df32c7 [PATCH] mm: ptd_a... |
2091 2092 2093 |
src_pte = huge_pte_offset(src, addr); if (!src_pte) continue; |
a55164389 hugetlb: modular ... |
2094 |
dst_pte = huge_pte_alloc(dst, addr, sz); |
63551ae0f [PATCH] Hugepage ... |
2095 2096 |
if (!dst_pte) goto nomem; |
c5c99429f fix hugepages lea... |
2097 2098 2099 2100 |
/* If the pagetables are shared don't copy or take references */ if (dst_pte == src_pte) continue; |
c74df32c7 [PATCH] mm: ptd_a... |
2101 |
spin_lock(&dst->page_table_lock); |
464787581 hugetlb: fix lock... |
2102 |
spin_lock_nested(&src->page_table_lock, SINGLE_DEPTH_NESTING); |
7f2e9525b hugetlbfs: common... |
2103 |
if (!huge_pte_none(huge_ptep_get(src_pte))) { |
1e8f889b1 [PATCH] Hugetlb: ... |
2104 |
if (cow) |
7f2e9525b hugetlbfs: common... |
2105 2106 |
huge_ptep_set_wrprotect(src, addr, src_pte); entry = huge_ptep_get(src_pte); |
1c59827d1 [PATCH] mm: huget... |
2107 2108 |
ptepage = pte_page(entry); get_page(ptepage); |
0fe6e20b9 hugetlb, rmap: ad... |
2109 |
page_dup_rmap(ptepage); |
1c59827d1 [PATCH] mm: huget... |
2110 2111 2112 |
set_huge_pte_at(dst, addr, dst_pte, entry); } spin_unlock(&src->page_table_lock); |
c74df32c7 [PATCH] mm: ptd_a... |
2113 |
spin_unlock(&dst->page_table_lock); |
63551ae0f [PATCH] Hugepage ... |
2114 2115 2116 2117 2118 2119 |
} return 0; nomem: return -ENOMEM; } |
290408d4a hugetlb: hugepage... |
2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 |
static int is_hugetlb_entry_migration(pte_t pte) { swp_entry_t swp; if (huge_pte_none(pte) || pte_present(pte)) return 0; swp = pte_to_swp_entry(pte); if (non_swap_entry(swp) && is_migration_entry(swp)) { return 1; } else return 0; } |
fd6a03edd HWPOISON, hugetlb... |
2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 |
static int is_hugetlb_entry_hwpoisoned(pte_t pte) { swp_entry_t swp; if (huge_pte_none(pte) || pte_present(pte)) return 0; swp = pte_to_swp_entry(pte); if (non_swap_entry(swp) && is_hwpoison_entry(swp)) { return 1; } else return 0; } |
502717f4e [PATCH] hugetlb: ... |
2144 |
void __unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start, |
04f2cbe35 hugetlb: guarante... |
2145 |
unsigned long end, struct page *ref_page) |
63551ae0f [PATCH] Hugepage ... |
2146 2147 2148 |
{ struct mm_struct *mm = vma->vm_mm; unsigned long address; |
c7546f8f0 [PATCH] Fix hugep... |
2149 |
pte_t *ptep; |
63551ae0f [PATCH] Hugepage ... |
2150 2151 |
pte_t pte; struct page *page; |
fe1668ae5 [PATCH] enforce p... |
2152 |
struct page *tmp; |
a55164389 hugetlb: modular ... |
2153 2154 |
struct hstate *h = hstate_vma(vma); unsigned long sz = huge_page_size(h); |
c0a499c2c [PATCH] __unmap_h... |
2155 2156 2157 2158 2159 |
/* * A page gathering list, protected by per file i_mmap_lock. The * lock is used to avoid list corruption from multiple unmapping * of the same page since we are using page->lru. */ |
fe1668ae5 [PATCH] enforce p... |
2160 |
LIST_HEAD(page_list); |
63551ae0f [PATCH] Hugepage ... |
2161 2162 |
WARN_ON(!is_vm_hugetlb_page(vma)); |
a55164389 hugetlb: modular ... |
2163 2164 |
BUG_ON(start & ~huge_page_mask(h)); BUG_ON(end & ~huge_page_mask(h)); |
63551ae0f [PATCH] Hugepage ... |
2165 |
|
cddb8a5c1 mmu-notifiers: core |
2166 |
mmu_notifier_invalidate_range_start(mm, start, end); |
508034a32 [PATCH] mm: unmap... |
2167 |
spin_lock(&mm->page_table_lock); |
a55164389 hugetlb: modular ... |
2168 |
for (address = start; address < end; address += sz) { |
c7546f8f0 [PATCH] Fix hugep... |
2169 |
ptep = huge_pte_offset(mm, address); |
4c8872659 [PATCH] hugetlb: ... |
2170 |
if (!ptep) |
c7546f8f0 [PATCH] Fix hugep... |
2171 |
continue; |
39dde65c9 [PATCH] shared pa... |
2172 2173 |
if (huge_pmd_unshare(mm, &address, ptep)) continue; |
04f2cbe35 hugetlb: guarante... |
2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 |
/* * If a reference page is supplied, it is because a specific * page is being unmapped, not a range. Ensure the page we * are about to unmap is the actual page of interest. */ if (ref_page) { pte = huge_ptep_get(ptep); if (huge_pte_none(pte)) continue; page = pte_page(pte); if (page != ref_page) continue; /* * Mark the VMA as having unmapped its page so that * future faults in this VMA will fail rather than * looking like data was lost */ set_vma_resv_flags(vma, HPAGE_RESV_UNMAPPED); } |
c7546f8f0 [PATCH] Fix hugep... |
2194 |
pte = huge_ptep_get_and_clear(mm, address, ptep); |
7f2e9525b hugetlbfs: common... |
2195 |
if (huge_pte_none(pte)) |
63551ae0f [PATCH] Hugepage ... |
2196 |
continue; |
c7546f8f0 [PATCH] Fix hugep... |
2197 |
|
fd6a03edd HWPOISON, hugetlb... |
2198 2199 2200 2201 2202 |
/* * HWPoisoned hugepage is already unmapped and dropped reference */ if (unlikely(is_hugetlb_entry_hwpoisoned(pte))) continue; |
63551ae0f [PATCH] Hugepage ... |
2203 |
page = pte_page(pte); |
6649a3863 [PATCH] hugetlb: ... |
2204 2205 |
if (pte_dirty(pte)) set_page_dirty(page); |
fe1668ae5 [PATCH] enforce p... |
2206 |
list_add(&page->lru, &page_list); |
63551ae0f [PATCH] Hugepage ... |
2207 |
} |
1da177e4c Linux-2.6.12-rc2 |
2208 |
spin_unlock(&mm->page_table_lock); |
508034a32 [PATCH] mm: unmap... |
2209 |
flush_tlb_range(vma, start, end); |
cddb8a5c1 mmu-notifiers: core |
2210 |
mmu_notifier_invalidate_range_end(mm, start, end); |
fe1668ae5 [PATCH] enforce p... |
2211 |
list_for_each_entry_safe(page, tmp, &page_list, lru) { |
0fe6e20b9 hugetlb, rmap: ad... |
2212 |
page_remove_rmap(page); |
fe1668ae5 [PATCH] enforce p... |
2213 2214 2215 |
list_del(&page->lru); put_page(page); } |
1da177e4c Linux-2.6.12-rc2 |
2216 |
} |
63551ae0f [PATCH] Hugepage ... |
2217 |
|
502717f4e [PATCH] hugetlb: ... |
2218 |
void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start, |
04f2cbe35 hugetlb: guarante... |
2219 |
unsigned long end, struct page *ref_page) |
502717f4e [PATCH] hugetlb: ... |
2220 |
{ |
a137e1cc6 hugetlbfs: per mo... |
2221 2222 2223 |
spin_lock(&vma->vm_file->f_mapping->i_mmap_lock); __unmap_hugepage_range(vma, start, end, ref_page); spin_unlock(&vma->vm_file->f_mapping->i_mmap_lock); |
502717f4e [PATCH] hugetlb: ... |
2224 |
} |
04f2cbe35 hugetlb: guarante... |
2225 2226 2227 2228 2229 2230 |
/* * This is called when the original mapper is failing to COW a MAP_PRIVATE * mappping it owns the reserve page for. The intention is to unmap the page * from other VMAs and let the children be SIGKILLed if they are faulting the * same region. */ |
2a4b3ded5 mm: hugetlb.c mak... |
2231 2232 |
static int unmap_ref_private(struct mm_struct *mm, struct vm_area_struct *vma, struct page *page, unsigned long address) |
04f2cbe35 hugetlb: guarante... |
2233 |
{ |
7526674de hugetlb: make unm... |
2234 |
struct hstate *h = hstate_vma(vma); |
04f2cbe35 hugetlb: guarante... |
2235 2236 2237 2238 2239 2240 2241 2242 2243 |
struct vm_area_struct *iter_vma; struct address_space *mapping; struct prio_tree_iter iter; pgoff_t pgoff; /* * vm_pgoff is in PAGE_SIZE units, hence the different calculation * from page cache lookup which is in HPAGE_SIZE units. */ |
7526674de hugetlb: make unm... |
2244 |
address = address & huge_page_mask(h); |
04f2cbe35 hugetlb: guarante... |
2245 2246 2247 |
pgoff = ((address - vma->vm_start) >> PAGE_SHIFT) + (vma->vm_pgoff >> PAGE_SHIFT); mapping = (struct address_space *)page_private(page); |
4eb2b1dcd hugetlb: acquire ... |
2248 2249 2250 2251 2252 2253 |
/* * 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 */ spin_lock(&mapping->i_mmap_lock); |
04f2cbe35 hugetlb: guarante... |
2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 |
vma_prio_tree_foreach(iter_vma, &iter, &mapping->i_mmap, pgoff, pgoff) { /* Do not unmap the current VMA */ if (iter_vma == vma) continue; /* * Unmap the page from other VMAs without their own reserves. * They get marked to be SIGKILLed if they fault in these * areas. This is because a future no-page fault on this VMA * could insert a zeroed page instead of the data existing * from the time of fork. This would look like data corruption */ if (!is_vma_resv_set(iter_vma, HPAGE_RESV_OWNER)) |
4eb2b1dcd hugetlb: acquire ... |
2267 |
__unmap_hugepage_range(iter_vma, |
7526674de hugetlb: make unm... |
2268 |
address, address + huge_page_size(h), |
04f2cbe35 hugetlb: guarante... |
2269 2270 |
page); } |
4eb2b1dcd hugetlb: acquire ... |
2271 |
spin_unlock(&mapping->i_mmap_lock); |
04f2cbe35 hugetlb: guarante... |
2272 2273 2274 |
return 1; } |
0fe6e20b9 hugetlb, rmap: ad... |
2275 2276 2277 |
/* * Hugetlb_cow() should be called with page lock of the original hugepage held. */ |
1e8f889b1 [PATCH] Hugetlb: ... |
2278 |
static int hugetlb_cow(struct mm_struct *mm, struct vm_area_struct *vma, |
04f2cbe35 hugetlb: guarante... |
2279 2280 |
unsigned long address, pte_t *ptep, pte_t pte, struct page *pagecache_page) |
1e8f889b1 [PATCH] Hugetlb: ... |
2281 |
{ |
a55164389 hugetlb: modular ... |
2282 |
struct hstate *h = hstate_vma(vma); |
1e8f889b1 [PATCH] Hugetlb: ... |
2283 |
struct page *old_page, *new_page; |
79ac6ba40 [PATCH] hugepage:... |
2284 |
int avoidcopy; |
04f2cbe35 hugetlb: guarante... |
2285 |
int outside_reserve = 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
2286 2287 |
old_page = pte_page(pte); |
04f2cbe35 hugetlb: guarante... |
2288 |
retry_avoidcopy: |
1e8f889b1 [PATCH] Hugetlb: ... |
2289 2290 |
/* If no-one else is actually using this page, avoid the copy * and just make the page writable */ |
0fe6e20b9 hugetlb, rmap: ad... |
2291 |
avoidcopy = (page_mapcount(old_page) == 1); |
1e8f889b1 [PATCH] Hugetlb: ... |
2292 |
if (avoidcopy) { |
56c9cfb13 hugetlb, rmap: fi... |
2293 2294 |
if (PageAnon(old_page)) page_move_anon_rmap(old_page, vma, address); |
1e8f889b1 [PATCH] Hugetlb: ... |
2295 |
set_huge_ptep_writable(vma, address, ptep); |
83c54070e mm: fault feedbac... |
2296 |
return 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
2297 |
} |
04f2cbe35 hugetlb: guarante... |
2298 2299 2300 2301 2302 2303 2304 2305 2306 |
/* * If the process that created a MAP_PRIVATE mapping is about to * perform a COW due to a shared page count, attempt to satisfy * the allocation without using the existing reserves. The pagecache * page is used to determine if the reserve at this address was * consumed or not. If reserves were used, a partial faulted mapping * at the time of fork() could consume its reserves on COW instead * of the full address range. */ |
f83a275db mm: account for M... |
2307 |
if (!(vma->vm_flags & VM_MAYSHARE) && |
04f2cbe35 hugetlb: guarante... |
2308 2309 2310 |
is_vma_resv_set(vma, HPAGE_RESV_OWNER) && old_page != pagecache_page) outside_reserve = 1; |
1e8f889b1 [PATCH] Hugetlb: ... |
2311 |
page_cache_get(old_page); |
b76c8cfbf hugetlb: prevent ... |
2312 2313 2314 |
/* Drop page_table_lock as buddy allocator may be called */ spin_unlock(&mm->page_table_lock); |
04f2cbe35 hugetlb: guarante... |
2315 |
new_page = alloc_huge_page(vma, address, outside_reserve); |
1e8f889b1 [PATCH] Hugetlb: ... |
2316 |
|
2fc39cec6 hugetlb: debit qu... |
2317 |
if (IS_ERR(new_page)) { |
1e8f889b1 [PATCH] Hugetlb: ... |
2318 |
page_cache_release(old_page); |
04f2cbe35 hugetlb: guarante... |
2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 |
/* * If a process owning a MAP_PRIVATE mapping fails to COW, * it is due to references held by a child and an insufficient * huge page pool. To guarantee the original mappers * reliability, unmap the page from child processes. The child * may get SIGKILLed if it later faults. */ if (outside_reserve) { BUG_ON(huge_pte_none(pte)); if (unmap_ref_private(mm, vma, old_page, address)) { BUG_ON(page_count(old_page) != 1); BUG_ON(huge_pte_none(pte)); |
b76c8cfbf hugetlb: prevent ... |
2332 |
spin_lock(&mm->page_table_lock); |
04f2cbe35 hugetlb: guarante... |
2333 2334 2335 2336 |
goto retry_avoidcopy; } WARN_ON_ONCE(1); } |
b76c8cfbf hugetlb: prevent ... |
2337 2338 |
/* Caller expects lock to be held */ spin_lock(&mm->page_table_lock); |
2fc39cec6 hugetlb: debit qu... |
2339 |
return -PTR_ERR(new_page); |
1e8f889b1 [PATCH] Hugetlb: ... |
2340 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
2341 2342 2343 2344 |
/* * When the original hugepage is shared one, it does not have * anon_vma prepared. */ |
44e2aa937 mm/hugetlb.c: add... |
2345 2346 2347 |
if (unlikely(anon_vma_prepare(vma))) { /* Caller expects lock to be held */ spin_lock(&mm->page_table_lock); |
0fe6e20b9 hugetlb, rmap: ad... |
2348 |
return VM_FAULT_OOM; |
44e2aa937 mm/hugetlb.c: add... |
2349 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
2350 |
|
0ebabb416 hugetlb: redefine... |
2351 |
copy_user_huge_page(new_page, old_page, address, vma); |
0ed361dec mm: fix PageUptod... |
2352 |
__SetPageUptodate(new_page); |
1e8f889b1 [PATCH] Hugetlb: ... |
2353 |
|
b76c8cfbf hugetlb: prevent ... |
2354 2355 2356 2357 2358 |
/* * Retake the page_table_lock to check for racing updates * before the page tables are altered */ spin_lock(&mm->page_table_lock); |
a55164389 hugetlb: modular ... |
2359 |
ptep = huge_pte_offset(mm, address & huge_page_mask(h)); |
7f2e9525b hugetlbfs: common... |
2360 |
if (likely(pte_same(huge_ptep_get(ptep), pte))) { |
1e8f889b1 [PATCH] Hugetlb: ... |
2361 |
/* Break COW */ |
3edd4fc95 hugetlb: call mmu... |
2362 2363 2364 |
mmu_notifier_invalidate_range_start(mm, address & huge_page_mask(h), (address & huge_page_mask(h)) + huge_page_size(h)); |
8fe627ec5 hugetlbfs: add mi... |
2365 |
huge_ptep_clear_flush(vma, address, ptep); |
1e8f889b1 [PATCH] Hugetlb: ... |
2366 2367 |
set_huge_pte_at(mm, address, ptep, make_huge_pte(vma, new_page, 1)); |
0fe6e20b9 hugetlb, rmap: ad... |
2368 |
page_remove_rmap(old_page); |
cd67f0d2a hugetlb, rmap: us... |
2369 |
hugepage_add_new_anon_rmap(new_page, vma, address); |
1e8f889b1 [PATCH] Hugetlb: ... |
2370 2371 |
/* Make the old page be freed below */ new_page = old_page; |
3edd4fc95 hugetlb: call mmu... |
2372 2373 2374 |
mmu_notifier_invalidate_range_end(mm, address & huge_page_mask(h), (address & huge_page_mask(h)) + huge_page_size(h)); |
1e8f889b1 [PATCH] Hugetlb: ... |
2375 2376 2377 |
} page_cache_release(new_page); page_cache_release(old_page); |
83c54070e mm: fault feedbac... |
2378 |
return 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
2379 |
} |
04f2cbe35 hugetlb: guarante... |
2380 |
/* Return the pagecache page at a given address within a VMA */ |
a55164389 hugetlb: modular ... |
2381 2382 |
static struct page *hugetlbfs_pagecache_page(struct hstate *h, struct vm_area_struct *vma, unsigned long address) |
04f2cbe35 hugetlb: guarante... |
2383 2384 |
{ struct address_space *mapping; |
e7c4b0bfd huge page private... |
2385 |
pgoff_t idx; |
04f2cbe35 hugetlb: guarante... |
2386 2387 |
mapping = vma->vm_file->f_mapping; |
a55164389 hugetlb: modular ... |
2388 |
idx = vma_hugecache_offset(h, vma, address); |
04f2cbe35 hugetlb: guarante... |
2389 2390 2391 |
return find_lock_page(mapping, idx); } |
3ae77f43b mm: hugetlbfs_pag... |
2392 2393 2394 2395 2396 |
/* * 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... |
2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 |
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... |
2411 |
static int hugetlb_no_page(struct mm_struct *mm, struct vm_area_struct *vma, |
788c7df45 hugetlb: fault fl... |
2412 |
unsigned long address, pte_t *ptep, unsigned int flags) |
ac9b9c667 [PATCH] Fix handl... |
2413 |
{ |
a55164389 hugetlb: modular ... |
2414 |
struct hstate *h = hstate_vma(vma); |
ac9b9c667 [PATCH] Fix handl... |
2415 |
int ret = VM_FAULT_SIGBUS; |
e7c4b0bfd huge page private... |
2416 |
pgoff_t idx; |
4c8872659 [PATCH] hugetlb: ... |
2417 |
unsigned long size; |
4c8872659 [PATCH] hugetlb: ... |
2418 2419 |
struct page *page; struct address_space *mapping; |
1e8f889b1 [PATCH] Hugetlb: ... |
2420 |
pte_t new_pte; |
4c8872659 [PATCH] hugetlb: ... |
2421 |
|
04f2cbe35 hugetlb: guarante... |
2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 |
/* * Currently, we are forced to kill the process in the event the * original mapper has unmapped pages from the child due to a failed * COW. Warn that such a situation has occured as it may not be obvious */ if (is_vma_resv_set(vma, HPAGE_RESV_UNMAPPED)) { printk(KERN_WARNING "PID %d killed due to inadequate hugepage pool ", current->pid); return ret; } |
4c8872659 [PATCH] hugetlb: ... |
2434 |
mapping = vma->vm_file->f_mapping; |
a55164389 hugetlb: modular ... |
2435 |
idx = vma_hugecache_offset(h, vma, address); |
4c8872659 [PATCH] hugetlb: ... |
2436 2437 2438 2439 2440 |
/* * Use page lock to guard against racing truncation * before we get page_table_lock. */ |
6bda666a0 [PATCH] hugepages... |
2441 2442 2443 |
retry: page = find_lock_page(mapping, idx); if (!page) { |
a55164389 hugetlb: modular ... |
2444 |
size = i_size_read(mapping->host) >> huge_page_shift(h); |
ebed4bfc8 [PATCH] hugetlb: ... |
2445 2446 |
if (idx >= size) goto out; |
04f2cbe35 hugetlb: guarante... |
2447 |
page = alloc_huge_page(vma, address, 0); |
2fc39cec6 hugetlb: debit qu... |
2448 2449 |
if (IS_ERR(page)) { ret = -PTR_ERR(page); |
6bda666a0 [PATCH] hugepages... |
2450 2451 |
goto out; } |
a55164389 hugetlb: modular ... |
2452 |
clear_huge_page(page, address, huge_page_size(h)); |
0ed361dec mm: fix PageUptod... |
2453 |
__SetPageUptodate(page); |
ac9b9c667 [PATCH] Fix handl... |
2454 |
|
f83a275db mm: account for M... |
2455 |
if (vma->vm_flags & VM_MAYSHARE) { |
6bda666a0 [PATCH] hugepages... |
2456 |
int err; |
45c682a68 hugetlb: fix i_bl... |
2457 |
struct inode *inode = mapping->host; |
6bda666a0 [PATCH] hugepages... |
2458 2459 2460 2461 |
err = add_to_page_cache(page, mapping, idx, GFP_KERNEL); if (err) { put_page(page); |
6bda666a0 [PATCH] hugepages... |
2462 2463 2464 2465 |
if (err == -EEXIST) goto retry; goto out; } |
45c682a68 hugetlb: fix i_bl... |
2466 2467 |
spin_lock(&inode->i_lock); |
a55164389 hugetlb: modular ... |
2468 |
inode->i_blocks += blocks_per_huge_page(h); |
45c682a68 hugetlb: fix i_bl... |
2469 |
spin_unlock(&inode->i_lock); |
0fe6e20b9 hugetlb, rmap: ad... |
2470 |
page_dup_rmap(page); |
23be7468e hugetlb: fix infi... |
2471 |
} else { |
6bda666a0 [PATCH] hugepages... |
2472 |
lock_page(page); |
0fe6e20b9 hugetlb, rmap: ad... |
2473 2474 2475 2476 2477 |
if (unlikely(anon_vma_prepare(vma))) { ret = VM_FAULT_OOM; goto backout_unlocked; } hugepage_add_new_anon_rmap(page, vma, address); |
23be7468e hugetlb: fix infi... |
2478 |
} |
0fe6e20b9 hugetlb, rmap: ad... |
2479 |
} else { |
998b4382c hugetlb: fix meta... |
2480 2481 2482 2483 2484 2485 |
/* * If memory error occurs between mmap() and fault, some process * don't have hwpoisoned swap entry for errored virtual address. * So we need to block hugepage fault by PG_hwpoison bit check. */ if (unlikely(PageHWPoison(page))) { |
aa50d3a7a Encode huge page ... |
2486 2487 |
ret = VM_FAULT_HWPOISON | VM_FAULT_SET_HINDEX(h - hstates); |
998b4382c hugetlb: fix meta... |
2488 2489 |
goto backout_unlocked; } |
0fe6e20b9 hugetlb, rmap: ad... |
2490 |
page_dup_rmap(page); |
6bda666a0 [PATCH] hugepages... |
2491 |
} |
1e8f889b1 [PATCH] Hugetlb: ... |
2492 |
|
57303d801 hugetlbfs: alloca... |
2493 2494 2495 2496 2497 2498 |
/* * 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... |
2499 |
if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) |
2b26736c8 allocate structur... |
2500 2501 2502 2503 |
if (vma_needs_reservation(h, vma, address) < 0) { ret = VM_FAULT_OOM; goto backout_unlocked; } |
57303d801 hugetlbfs: alloca... |
2504 |
|
ac9b9c667 [PATCH] Fix handl... |
2505 |
spin_lock(&mm->page_table_lock); |
a55164389 hugetlb: modular ... |
2506 |
size = i_size_read(mapping->host) >> huge_page_shift(h); |
4c8872659 [PATCH] hugetlb: ... |
2507 2508 |
if (idx >= size) goto backout; |
83c54070e mm: fault feedbac... |
2509 |
ret = 0; |
7f2e9525b hugetlbfs: common... |
2510 |
if (!huge_pte_none(huge_ptep_get(ptep))) |
4c8872659 [PATCH] hugetlb: ... |
2511 |
goto backout; |
1e8f889b1 [PATCH] Hugetlb: ... |
2512 2513 2514 |
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... |
2515 |
if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) { |
1e8f889b1 [PATCH] Hugetlb: ... |
2516 |
/* Optimization, do the COW without a second fault */ |
04f2cbe35 hugetlb: guarante... |
2517 |
ret = hugetlb_cow(mm, vma, address, ptep, new_pte, page); |
1e8f889b1 [PATCH] Hugetlb: ... |
2518 |
} |
ac9b9c667 [PATCH] Fix handl... |
2519 |
spin_unlock(&mm->page_table_lock); |
4c8872659 [PATCH] hugetlb: ... |
2520 2521 |
unlock_page(page); out: |
ac9b9c667 [PATCH] Fix handl... |
2522 |
return ret; |
4c8872659 [PATCH] hugetlb: ... |
2523 2524 2525 |
backout: spin_unlock(&mm->page_table_lock); |
2b26736c8 allocate structur... |
2526 |
backout_unlocked: |
4c8872659 [PATCH] hugetlb: ... |
2527 2528 2529 |
unlock_page(page); put_page(page); goto out; |
ac9b9c667 [PATCH] Fix handl... |
2530 |
} |
86e5216f8 [PATCH] Hugetlb: ... |
2531 |
int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
788c7df45 hugetlb: fault fl... |
2532 |
unsigned long address, unsigned int flags) |
86e5216f8 [PATCH] Hugetlb: ... |
2533 2534 2535 |
{ pte_t *ptep; pte_t entry; |
1e8f889b1 [PATCH] Hugetlb: ... |
2536 |
int ret; |
0fe6e20b9 hugetlb, rmap: ad... |
2537 |
struct page *page = NULL; |
57303d801 hugetlbfs: alloca... |
2538 |
struct page *pagecache_page = NULL; |
3935baa9b [PATCH] hugepage:... |
2539 |
static DEFINE_MUTEX(hugetlb_instantiation_mutex); |
a55164389 hugetlb: modular ... |
2540 |
struct hstate *h = hstate_vma(vma); |
86e5216f8 [PATCH] Hugetlb: ... |
2541 |
|
fd6a03edd HWPOISON, hugetlb... |
2542 2543 2544 |
ptep = huge_pte_offset(mm, address); if (ptep) { entry = huge_ptep_get(ptep); |
290408d4a hugetlb: hugepage... |
2545 2546 2547 2548 |
if (unlikely(is_hugetlb_entry_migration(entry))) { migration_entry_wait(mm, (pmd_t *)ptep, address); return 0; } else if (unlikely(is_hugetlb_entry_hwpoisoned(entry))) |
aa50d3a7a Encode huge page ... |
2549 2550 |
return VM_FAULT_HWPOISON_LARGE | VM_FAULT_SET_HINDEX(h - hstates); |
fd6a03edd HWPOISON, hugetlb... |
2551 |
} |
a55164389 hugetlb: modular ... |
2552 |
ptep = huge_pte_alloc(mm, address, huge_page_size(h)); |
86e5216f8 [PATCH] Hugetlb: ... |
2553 2554 |
if (!ptep) return VM_FAULT_OOM; |
3935baa9b [PATCH] hugepage:... |
2555 2556 2557 2558 2559 2560 |
/* * Serialize hugepage allocation and instantiation, so that we don't * get spurious allocation failures if two CPUs race to instantiate * the same page in the page cache. */ mutex_lock(&hugetlb_instantiation_mutex); |
7f2e9525b hugetlbfs: common... |
2561 2562 |
entry = huge_ptep_get(ptep); if (huge_pte_none(entry)) { |
788c7df45 hugetlb: fault fl... |
2563 |
ret = hugetlb_no_page(mm, vma, address, ptep, flags); |
b4d1d99fd hugetlb: handle u... |
2564 |
goto out_mutex; |
3935baa9b [PATCH] hugepage:... |
2565 |
} |
86e5216f8 [PATCH] Hugetlb: ... |
2566 |
|
83c54070e mm: fault feedbac... |
2567 |
ret = 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
2568 |
|
57303d801 hugetlbfs: alloca... |
2569 2570 2571 2572 2573 2574 2575 2576 |
/* * If we are going to COW the mapping later, we examine the pending * reservations for this page now. This will ensure that any * allocations necessary to record that reservation occur outside the * spinlock. For private mappings, we also lookup the pagecache * page now as it is used to determine if a reservation has been * consumed. */ |
788c7df45 hugetlb: fault fl... |
2577 |
if ((flags & FAULT_FLAG_WRITE) && !pte_write(entry)) { |
2b26736c8 allocate structur... |
2578 2579 |
if (vma_needs_reservation(h, vma, address) < 0) { ret = VM_FAULT_OOM; |
b4d1d99fd hugetlb: handle u... |
2580 |
goto out_mutex; |
2b26736c8 allocate structur... |
2581 |
} |
57303d801 hugetlbfs: alloca... |
2582 |
|
f83a275db mm: account for M... |
2583 |
if (!(vma->vm_flags & VM_MAYSHARE)) |
57303d801 hugetlbfs: alloca... |
2584 2585 2586 |
pagecache_page = hugetlbfs_pagecache_page(h, vma, address); } |
56c9cfb13 hugetlb, rmap: fi... |
2587 2588 2589 2590 2591 2592 2593 2594 2595 |
/* * hugetlb_cow() requires page locks of pte_page(entry) and * pagecache_page, so here we need take the former one * when page != pagecache_page or !pagecache_page. * Note that locking order is always pagecache_page -> page, * so no worry about deadlock. */ page = pte_page(entry); if (page != pagecache_page) |
0fe6e20b9 hugetlb, rmap: ad... |
2596 |
lock_page(page); |
0fe6e20b9 hugetlb, rmap: ad... |
2597 |
|
1e8f889b1 [PATCH] Hugetlb: ... |
2598 2599 |
spin_lock(&mm->page_table_lock); /* Check for a racing update before calling hugetlb_cow */ |
b4d1d99fd hugetlb: handle u... |
2600 2601 |
if (unlikely(!pte_same(entry, huge_ptep_get(ptep)))) goto out_page_table_lock; |
788c7df45 hugetlb: fault fl... |
2602 |
if (flags & FAULT_FLAG_WRITE) { |
b4d1d99fd hugetlb: handle u... |
2603 |
if (!pte_write(entry)) { |
57303d801 hugetlbfs: alloca... |
2604 2605 |
ret = hugetlb_cow(mm, vma, address, ptep, entry, pagecache_page); |
b4d1d99fd hugetlb: handle u... |
2606 2607 2608 2609 2610 |
goto out_page_table_lock; } entry = pte_mkdirty(entry); } entry = pte_mkyoung(entry); |
788c7df45 hugetlb: fault fl... |
2611 2612 |
if (huge_ptep_set_access_flags(vma, address, ptep, entry, flags & FAULT_FLAG_WRITE)) |
4b3073e1c MM: Pass a PTE po... |
2613 |
update_mmu_cache(vma, address, ptep); |
b4d1d99fd hugetlb: handle u... |
2614 2615 |
out_page_table_lock: |
1e8f889b1 [PATCH] Hugetlb: ... |
2616 |
spin_unlock(&mm->page_table_lock); |
57303d801 hugetlbfs: alloca... |
2617 2618 2619 2620 2621 |
if (pagecache_page) { unlock_page(pagecache_page); put_page(pagecache_page); } |
1f64d69c7 mm/hugetlb.c: avo... |
2622 2623 |
if (page != pagecache_page) unlock_page(page); |
57303d801 hugetlbfs: alloca... |
2624 |
|
b4d1d99fd hugetlb: handle u... |
2625 |
out_mutex: |
3935baa9b [PATCH] hugepage:... |
2626 |
mutex_unlock(&hugetlb_instantiation_mutex); |
1e8f889b1 [PATCH] Hugetlb: ... |
2627 2628 |
return ret; |
86e5216f8 [PATCH] Hugetlb: ... |
2629 |
} |
ceb868796 hugetlb: introduc... |
2630 2631 2632 2633 2634 2635 2636 2637 |
/* Can be overriden by architectures */ __attribute__((weak)) struct page * follow_huge_pud(struct mm_struct *mm, unsigned long address, pud_t *pud, int write) { BUG(); return NULL; } |
63551ae0f [PATCH] Hugepage ... |
2638 2639 |
int follow_hugetlb_page(struct mm_struct *mm, struct vm_area_struct *vma, struct page **pages, struct vm_area_struct **vmas, |
5b23dbe81 hugetlb: follow_h... |
2640 |
unsigned long *position, int *length, int i, |
2a15efc95 mm: follow_hugetl... |
2641 |
unsigned int flags) |
63551ae0f [PATCH] Hugepage ... |
2642 |
{ |
d5d4b0aa4 [PATCH] optimize ... |
2643 2644 |
unsigned long pfn_offset; unsigned long vaddr = *position; |
63551ae0f [PATCH] Hugepage ... |
2645 |
int remainder = *length; |
a55164389 hugetlb: modular ... |
2646 |
struct hstate *h = hstate_vma(vma); |
63551ae0f [PATCH] Hugepage ... |
2647 |
|
1c59827d1 [PATCH] mm: huget... |
2648 |
spin_lock(&mm->page_table_lock); |
63551ae0f [PATCH] Hugepage ... |
2649 |
while (vaddr < vma->vm_end && remainder) { |
4c8872659 [PATCH] hugetlb: ... |
2650 |
pte_t *pte; |
2a15efc95 mm: follow_hugetl... |
2651 |
int absent; |
4c8872659 [PATCH] hugetlb: ... |
2652 |
struct page *page; |
63551ae0f [PATCH] Hugepage ... |
2653 |
|
4c8872659 [PATCH] hugetlb: ... |
2654 2655 |
/* * Some archs (sparc64, sh*) have multiple pte_ts to |
2a15efc95 mm: follow_hugetl... |
2656 |
* each hugepage. We have to make sure we get the |
4c8872659 [PATCH] hugetlb: ... |
2657 2658 |
* first, for the page indexing below to work. */ |
a55164389 hugetlb: modular ... |
2659 |
pte = huge_pte_offset(mm, vaddr & huge_page_mask(h)); |
2a15efc95 mm: follow_hugetl... |
2660 2661 2662 2663 |
absent = !pte || huge_pte_none(huge_ptep_get(pte)); /* * When coredumping, it suits get_dump_page if we just return |
3ae77f43b mm: hugetlbfs_pag... |
2664 2665 2666 2667 |
* 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... |
2668 |
*/ |
3ae77f43b mm: hugetlbfs_pag... |
2669 2670 |
if (absent && (flags & FOLL_DUMP) && !hugetlbfs_pagecache_present(h, vma, vaddr)) { |
2a15efc95 mm: follow_hugetl... |
2671 2672 2673 |
remainder = 0; break; } |
63551ae0f [PATCH] Hugepage ... |
2674 |
|
2a15efc95 mm: follow_hugetl... |
2675 2676 |
if (absent || ((flags & FOLL_WRITE) && !pte_write(huge_ptep_get(pte)))) { |
4c8872659 [PATCH] hugetlb: ... |
2677 |
int ret; |
63551ae0f [PATCH] Hugepage ... |
2678 |
|
4c8872659 [PATCH] hugetlb: ... |
2679 |
spin_unlock(&mm->page_table_lock); |
2a15efc95 mm: follow_hugetl... |
2680 2681 |
ret = hugetlb_fault(mm, vma, vaddr, (flags & FOLL_WRITE) ? FAULT_FLAG_WRITE : 0); |
4c8872659 [PATCH] hugetlb: ... |
2682 |
spin_lock(&mm->page_table_lock); |
a89182c76 Fix VM_FAULT flag... |
2683 |
if (!(ret & VM_FAULT_ERROR)) |
4c8872659 [PATCH] hugetlb: ... |
2684 |
continue; |
63551ae0f [PATCH] Hugepage ... |
2685 |
|
4c8872659 [PATCH] hugetlb: ... |
2686 |
remainder = 0; |
4c8872659 [PATCH] hugetlb: ... |
2687 2688 |
break; } |
a55164389 hugetlb: modular ... |
2689 |
pfn_offset = (vaddr & ~huge_page_mask(h)) >> PAGE_SHIFT; |
7f2e9525b hugetlbfs: common... |
2690 |
page = pte_page(huge_ptep_get(pte)); |
d5d4b0aa4 [PATCH] optimize ... |
2691 |
same_page: |
d6692183a [PATCH] fix extra... |
2692 |
if (pages) { |
2a15efc95 mm: follow_hugetl... |
2693 |
pages[i] = mem_map_offset(page, pfn_offset); |
4b2e38ad7 hugepage: support... |
2694 |
get_page(pages[i]); |
d6692183a [PATCH] fix extra... |
2695 |
} |
63551ae0f [PATCH] Hugepage ... |
2696 2697 2698 2699 2700 |
if (vmas) vmas[i] = vma; vaddr += PAGE_SIZE; |
d5d4b0aa4 [PATCH] optimize ... |
2701 |
++pfn_offset; |
63551ae0f [PATCH] Hugepage ... |
2702 2703 |
--remainder; ++i; |
d5d4b0aa4 [PATCH] optimize ... |
2704 |
if (vaddr < vma->vm_end && remainder && |
a55164389 hugetlb: modular ... |
2705 |
pfn_offset < pages_per_huge_page(h)) { |
d5d4b0aa4 [PATCH] optimize ... |
2706 2707 2708 2709 2710 2711 |
/* * We use pfn_offset to avoid touching the pageframes * of this compound page. */ goto same_page; } |
63551ae0f [PATCH] Hugepage ... |
2712 |
} |
1c59827d1 [PATCH] mm: huget... |
2713 |
spin_unlock(&mm->page_table_lock); |
63551ae0f [PATCH] Hugepage ... |
2714 2715 |
*length = remainder; *position = vaddr; |
2a15efc95 mm: follow_hugetl... |
2716 |
return i ? i : -EFAULT; |
63551ae0f [PATCH] Hugepage ... |
2717 |
} |
8f860591f [PATCH] Enable mp... |
2718 2719 2720 2721 2722 2723 2724 2725 |
void hugetlb_change_protection(struct vm_area_struct *vma, unsigned long address, unsigned long end, pgprot_t newprot) { struct mm_struct *mm = vma->vm_mm; unsigned long start = address; pte_t *ptep; pte_t pte; |
a55164389 hugetlb: modular ... |
2726 |
struct hstate *h = hstate_vma(vma); |
8f860591f [PATCH] Enable mp... |
2727 2728 2729 |
BUG_ON(address >= end); flush_cache_range(vma, address, end); |
39dde65c9 [PATCH] shared pa... |
2730 |
spin_lock(&vma->vm_file->f_mapping->i_mmap_lock); |
8f860591f [PATCH] Enable mp... |
2731 |
spin_lock(&mm->page_table_lock); |
a55164389 hugetlb: modular ... |
2732 |
for (; address < end; address += huge_page_size(h)) { |
8f860591f [PATCH] Enable mp... |
2733 2734 2735 |
ptep = huge_pte_offset(mm, address); if (!ptep) continue; |
39dde65c9 [PATCH] shared pa... |
2736 2737 |
if (huge_pmd_unshare(mm, &address, ptep)) continue; |
7f2e9525b hugetlbfs: common... |
2738 |
if (!huge_pte_none(huge_ptep_get(ptep))) { |
8f860591f [PATCH] Enable mp... |
2739 2740 2741 |
pte = huge_ptep_get_and_clear(mm, address, ptep); pte = pte_mkhuge(pte_modify(pte, newprot)); set_huge_pte_at(mm, address, ptep, pte); |
8f860591f [PATCH] Enable mp... |
2742 2743 2744 |
} } spin_unlock(&mm->page_table_lock); |
39dde65c9 [PATCH] shared pa... |
2745 |
spin_unlock(&vma->vm_file->f_mapping->i_mmap_lock); |
8f860591f [PATCH] Enable mp... |
2746 2747 2748 |
flush_tlb_range(vma, start, end); } |
a1e78772d hugetlb: reserve ... |
2749 2750 |
int hugetlb_reserve_pages(struct inode *inode, long from, long to, |
5a6fe1259 Do not account fo... |
2751 2752 |
struct vm_area_struct *vma, int acctflag) |
e4e574b76 hugetlb: Try to g... |
2753 |
{ |
17c9d12e1 Do not account fo... |
2754 |
long ret, chg; |
a55164389 hugetlb: modular ... |
2755 |
struct hstate *h = hstate_inode(inode); |
e4e574b76 hugetlb: Try to g... |
2756 |
|
a1e78772d hugetlb: reserve ... |
2757 |
/* |
17c9d12e1 Do not account fo... |
2758 2759 2760 2761 2762 2763 2764 2765 |
* Only apply hugepage reservation if asked. At fault time, an * attempt will be made for VM_NORESERVE to allocate a page * and filesystem quota without using reserves */ if (acctflag & VM_NORESERVE) return 0; /* |
a1e78772d hugetlb: reserve ... |
2766 2767 2768 2769 2770 |
* Shared mappings base their reservation on the number of pages that * are already allocated on behalf of the file. Private mappings need * to reserve the full area even if read-only as mprotect() may be * called to make the mapping read-write. Assume !vma is a shm mapping */ |
f83a275db mm: account for M... |
2771 |
if (!vma || vma->vm_flags & VM_MAYSHARE) |
a1e78772d hugetlb: reserve ... |
2772 |
chg = region_chg(&inode->i_mapping->private_list, from, to); |
17c9d12e1 Do not account fo... |
2773 2774 2775 2776 |
else { struct resv_map *resv_map = resv_map_alloc(); if (!resv_map) return -ENOMEM; |
a1e78772d hugetlb: reserve ... |
2777 |
chg = to - from; |
84afd99b8 hugetlb reservati... |
2778 |
|
17c9d12e1 Do not account fo... |
2779 2780 2781 |
set_vma_resv_map(vma, resv_map); set_vma_resv_flags(vma, HPAGE_RESV_OWNER); } |
e4e574b76 hugetlb: Try to g... |
2782 2783 |
if (chg < 0) return chg; |
8a6301127 pretend cpuset ha... |
2784 |
|
17c9d12e1 Do not account fo... |
2785 |
/* There must be enough filesystem quota for the mapping */ |
90d8b7e61 hugetlb: enforce ... |
2786 2787 |
if (hugetlb_get_quota(inode->i_mapping, chg)) return -ENOSPC; |
5a6fe1259 Do not account fo... |
2788 2789 |
/* |
17c9d12e1 Do not account fo... |
2790 2791 |
* Check enough hugepages are available for the reservation. * Hand back the quota if there are not |
5a6fe1259 Do not account fo... |
2792 |
*/ |
a55164389 hugetlb: modular ... |
2793 |
ret = hugetlb_acct_memory(h, chg); |
68842c9b9 hugetlbfs: fix qu... |
2794 2795 |
if (ret < 0) { hugetlb_put_quota(inode->i_mapping, chg); |
a43a8c39b [PATCH] tightenin... |
2796 |
return ret; |
68842c9b9 hugetlbfs: fix qu... |
2797 |
} |
17c9d12e1 Do not account fo... |
2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 |
/* * 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... |
2810 |
if (!vma || vma->vm_flags & VM_MAYSHARE) |
a1e78772d hugetlb: reserve ... |
2811 |
region_add(&inode->i_mapping->private_list, from, to); |
a43a8c39b [PATCH] tightenin... |
2812 2813 2814 2815 2816 |
return 0; } void hugetlb_unreserve_pages(struct inode *inode, long offset, long freed) { |
a55164389 hugetlb: modular ... |
2817 |
struct hstate *h = hstate_inode(inode); |
a43a8c39b [PATCH] tightenin... |
2818 |
long chg = region_truncate(&inode->i_mapping->private_list, offset); |
45c682a68 hugetlb: fix i_bl... |
2819 2820 |
spin_lock(&inode->i_lock); |
e4c6f8bed hugetlbfs: fix i_... |
2821 |
inode->i_blocks -= (blocks_per_huge_page(h) * freed); |
45c682a68 hugetlb: fix i_bl... |
2822 |
spin_unlock(&inode->i_lock); |
90d8b7e61 hugetlb: enforce ... |
2823 |
hugetlb_put_quota(inode->i_mapping, (chg - freed)); |
a55164389 hugetlb: modular ... |
2824 |
hugetlb_acct_memory(h, -(chg - freed)); |
a43a8c39b [PATCH] tightenin... |
2825 |
} |
93f70f900 HWPOISON, hugetlb... |
2826 |
|
d5bd91069 hugepage: move is... |
2827 |
#ifdef CONFIG_MEMORY_FAILURE |
6de2b1aab HWPOISON, hugetlb... |
2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 |
/* Should be called in hugetlb_lock */ static int is_hugepage_on_freelist(struct page *hpage) { struct page *page; struct page *tmp; struct hstate *h = page_hstate(hpage); int nid = page_to_nid(hpage); list_for_each_entry_safe(page, tmp, &h->hugepage_freelists[nid], lru) if (page == hpage) return 1; return 0; } |
93f70f900 HWPOISON, hugetlb... |
2841 2842 2843 2844 |
/* * This function is called from memory failure code. * Assume the caller holds page lock of the head page. */ |
6de2b1aab HWPOISON, hugetlb... |
2845 |
int dequeue_hwpoisoned_huge_page(struct page *hpage) |
93f70f900 HWPOISON, hugetlb... |
2846 2847 2848 |
{ struct hstate *h = page_hstate(hpage); int nid = page_to_nid(hpage); |
6de2b1aab HWPOISON, hugetlb... |
2849 |
int ret = -EBUSY; |
93f70f900 HWPOISON, hugetlb... |
2850 2851 |
spin_lock(&hugetlb_lock); |
6de2b1aab HWPOISON, hugetlb... |
2852 2853 |
if (is_hugepage_on_freelist(hpage)) { list_del(&hpage->lru); |
8c6c2ecb4 HWPOSION, hugetlb... |
2854 |
set_page_refcounted(hpage); |
6de2b1aab HWPOISON, hugetlb... |
2855 2856 2857 2858 |
h->free_huge_pages--; h->free_huge_pages_node[nid]--; ret = 0; } |
93f70f900 HWPOISON, hugetlb... |
2859 |
spin_unlock(&hugetlb_lock); |
6de2b1aab HWPOISON, hugetlb... |
2860 |
return ret; |
93f70f900 HWPOISON, hugetlb... |
2861 |
} |
6de2b1aab HWPOISON, hugetlb... |
2862 |
#endif |