Blame view
mm/hugetlb.c
33.6 KB
1da177e4c Linux-2.6.12-rc2 |
1 2 3 4 5 6 7 8 9 |
/* * Generic hugetlb support. * (C) William Irwin, April 2004 */ #include <linux/gfp.h> #include <linux/list.h> #include <linux/init.h> #include <linux/module.h> #include <linux/mm.h> |
1da177e4c Linux-2.6.12-rc2 |
10 11 12 |
#include <linux/sysctl.h> #include <linux/highmem.h> #include <linux/nodemask.h> |
63551ae0f [PATCH] Hugepage ... |
13 |
#include <linux/pagemap.h> |
5da7ca860 [PATCH] Add NUMA ... |
14 |
#include <linux/mempolicy.h> |
aea47ff36 [PATCH] mm: make ... |
15 |
#include <linux/cpuset.h> |
3935baa9b [PATCH] hugepage:... |
16 |
#include <linux/mutex.h> |
5da7ca860 [PATCH] Add NUMA ... |
17 |
|
63551ae0f [PATCH] Hugepage ... |
18 19 20 21 |
#include <asm/page.h> #include <asm/pgtable.h> #include <linux/hugetlb.h> |
7835e98b2 [PATCH] remove se... |
22 |
#include "internal.h" |
1da177e4c Linux-2.6.12-rc2 |
23 24 |
const unsigned long hugetlb_zero = 0, hugetlb_infinity = ~0UL; |
a43a8c39b [PATCH] tightenin... |
25 |
static unsigned long nr_huge_pages, free_huge_pages, resv_huge_pages; |
7893d1d50 hugetlb: Try to g... |
26 |
static unsigned long surplus_huge_pages; |
064d9efe9 hugetlb: fix over... |
27 |
static unsigned long nr_overcommit_huge_pages; |
1da177e4c Linux-2.6.12-rc2 |
28 |
unsigned long max_huge_pages; |
064d9efe9 hugetlb: fix over... |
29 |
unsigned long sysctl_overcommit_huge_pages; |
1da177e4c Linux-2.6.12-rc2 |
30 31 32 |
static struct list_head hugepage_freelists[MAX_NUMNODES]; static unsigned int nr_huge_pages_node[MAX_NUMNODES]; static unsigned int free_huge_pages_node[MAX_NUMNODES]; |
7893d1d50 hugetlb: Try to g... |
33 |
static unsigned int surplus_huge_pages_node[MAX_NUMNODES]; |
396faf030 Allow huge page a... |
34 35 |
static gfp_t htlb_alloc_mask = GFP_HIGHUSER; unsigned long hugepages_treat_as_movable; |
63b4613c3 hugetlb: fix huge... |
36 |
static int hugetlb_next_nid; |
396faf030 Allow huge page a... |
37 |
|
3935baa9b [PATCH] hugepage:... |
38 39 40 41 |
/* * Protects updates to hugepage_freelists, nr_huge_pages, and free_huge_pages */ static DEFINE_SPINLOCK(hugetlb_lock); |
0bd0f9fb1 [PATCH] hugetlb: ... |
42 |
|
79ac6ba40 [PATCH] hugepage:... |
43 44 45 46 47 48 49 |
static void clear_huge_page(struct page *page, unsigned long addr) { int i; might_sleep(); for (i = 0; i < (HPAGE_SIZE/PAGE_SIZE); i++) { cond_resched(); |
281e0e3b3 hugetlb: fix clea... |
50 |
clear_user_highpage(page + i, addr + i * PAGE_SIZE); |
79ac6ba40 [PATCH] hugepage:... |
51 52 53 54 |
} } static void copy_huge_page(struct page *dst, struct page *src, |
9de455b20 [PATCH] Pass vma ... |
55 |
unsigned long addr, struct vm_area_struct *vma) |
79ac6ba40 [PATCH] hugepage:... |
56 57 58 59 60 61 |
{ int i; might_sleep(); for (i = 0; i < HPAGE_SIZE/PAGE_SIZE; i++) { cond_resched(); |
9de455b20 [PATCH] Pass vma ... |
62 |
copy_user_highpage(dst + i, src + i, addr + i*PAGE_SIZE, vma); |
79ac6ba40 [PATCH] hugepage:... |
63 64 |
} } |
1da177e4c Linux-2.6.12-rc2 |
65 66 67 68 69 70 71 |
static void enqueue_huge_page(struct page *page) { int nid = page_to_nid(page); list_add(&page->lru, &hugepage_freelists[nid]); free_huge_pages++; free_huge_pages_node[nid]++; } |
348e1e04b hugetlb: fix pool... |
72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 |
static struct page *dequeue_huge_page(void) { int nid; struct page *page = NULL; for (nid = 0; nid < MAX_NUMNODES; ++nid) { if (!list_empty(&hugepage_freelists[nid])) { page = list_entry(hugepage_freelists[nid].next, struct page, lru); list_del(&page->lru); free_huge_pages--; free_huge_pages_node[nid]--; break; } } return page; } static struct page *dequeue_huge_page_vma(struct vm_area_struct *vma, |
5da7ca860 [PATCH] Add NUMA ... |
91 |
unsigned long address) |
1da177e4c Linux-2.6.12-rc2 |
92 |
{ |
31a5c6e4f hugetlb: remove u... |
93 |
int nid; |
1da177e4c Linux-2.6.12-rc2 |
94 |
struct page *page = NULL; |
480eccf9a Fix NUMA Memory P... |
95 |
struct mempolicy *mpol; |
19770b326 mm: filter based ... |
96 |
nodemask_t *nodemask; |
396faf030 Allow huge page a... |
97 |
struct zonelist *zonelist = huge_zonelist(vma, address, |
19770b326 mm: filter based ... |
98 |
htlb_alloc_mask, &mpol, &nodemask); |
dd1a239f6 mm: have zonelist... |
99 100 |
struct zone *zone; struct zoneref *z; |
1da177e4c Linux-2.6.12-rc2 |
101 |
|
19770b326 mm: filter based ... |
102 103 |
for_each_zone_zonelist_nodemask(zone, z, zonelist, MAX_NR_ZONES - 1, nodemask) { |
54a6eb5c4 mm: use two zonel... |
104 105 |
nid = zone_to_nid(zone); if (cpuset_zone_allowed_softwall(zone, htlb_alloc_mask) && |
3abf7afd4 dequeue_huge_page... |
106 107 108 109 110 111 |
!list_empty(&hugepage_freelists[nid])) { page = list_entry(hugepage_freelists[nid].next, struct page, lru); list_del(&page->lru); free_huge_pages--; free_huge_pages_node[nid]--; |
e4e574b76 hugetlb: Try to g... |
112 113 |
if (vma && vma->vm_flags & VM_MAYSHARE) resv_huge_pages--; |
5ab3ee7b1 fix hugetlb page ... |
114 |
break; |
3abf7afd4 dequeue_huge_page... |
115 |
} |
1da177e4c Linux-2.6.12-rc2 |
116 |
} |
52cd3b074 mempolicy: rework... |
117 |
mpol_cond_put(mpol); |
1da177e4c Linux-2.6.12-rc2 |
118 119 |
return page; } |
6af2acb66 hugetlb: Move upd... |
120 121 122 123 124 125 126 127 128 129 130 131 |
static void update_and_free_page(struct page *page) { int i; nr_huge_pages--; nr_huge_pages_node[page_to_nid(page)]--; for (i = 0; i < (HPAGE_SIZE / PAGE_SIZE); i++) { 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... |
132 |
arch_release_hugepage(page); |
6af2acb66 hugetlb: Move upd... |
133 134 |
__free_pages(page, HUGETLB_PAGE_ORDER); } |
27a85ef1b [PATCH] hugepage:... |
135 136 |
static void free_huge_page(struct page *page) { |
7893d1d50 hugetlb: Try to g... |
137 |
int nid = page_to_nid(page); |
c79fb75e5 hugetlb: fix quot... |
138 |
struct address_space *mapping; |
27a85ef1b [PATCH] hugepage:... |
139 |
|
c79fb75e5 hugetlb: fix quot... |
140 |
mapping = (struct address_space *) page_private(page); |
e5df70ab1 hugetlb: ensure w... |
141 |
set_page_private(page, 0); |
7893d1d50 hugetlb: Try to g... |
142 |
BUG_ON(page_count(page)); |
27a85ef1b [PATCH] hugepage:... |
143 144 145 |
INIT_LIST_HEAD(&page->lru); spin_lock(&hugetlb_lock); |
7893d1d50 hugetlb: Try to g... |
146 147 148 149 150 151 152 |
if (surplus_huge_pages_node[nid]) { update_and_free_page(page); surplus_huge_pages--; surplus_huge_pages_node[nid]--; } else { enqueue_huge_page(page); } |
27a85ef1b [PATCH] hugepage:... |
153 |
spin_unlock(&hugetlb_lock); |
c79fb75e5 hugetlb: fix quot... |
154 |
if (mapping) |
9a119c056 hugetlb: allow bu... |
155 |
hugetlb_put_quota(mapping, 1); |
27a85ef1b [PATCH] hugepage:... |
156 |
} |
7893d1d50 hugetlb: Try to g... |
157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 |
/* * 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. */ static int adjust_pool_surplus(int delta) { static int prev_nid; int nid = prev_nid; int ret = 0; VM_BUG_ON(delta != -1 && delta != 1); do { nid = next_node(nid, node_online_map); if (nid == MAX_NUMNODES) nid = first_node(node_online_map); /* To shrink on this node, there must be a surplus page */ if (delta < 0 && !surplus_huge_pages_node[nid]) continue; /* Surplus cannot exceed the total number of pages */ if (delta > 0 && surplus_huge_pages_node[nid] >= nr_huge_pages_node[nid]) continue; surplus_huge_pages += delta; surplus_huge_pages_node[nid] += delta; ret = 1; break; } while (nid != prev_nid); prev_nid = nid; return ret; } |
63b4613c3 hugetlb: fix huge... |
191 |
static struct page *alloc_fresh_huge_page_node(int nid) |
1da177e4c Linux-2.6.12-rc2 |
192 |
{ |
1da177e4c Linux-2.6.12-rc2 |
193 |
struct page *page; |
f96efd585 hugetlb: fix race... |
194 |
|
63b4613c3 hugetlb: fix huge... |
195 |
page = alloc_pages_node(nid, |
551883ae8 page allocator: e... |
196 197 |
htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE| __GFP_REPEAT|__GFP_NOWARN, |
63b4613c3 hugetlb: fix huge... |
198 |
HUGETLB_PAGE_ORDER); |
1da177e4c Linux-2.6.12-rc2 |
199 |
if (page) { |
7f2e9525b hugetlbfs: common... |
200 201 |
if (arch_prepare_hugepage(page)) { __free_pages(page, HUGETLB_PAGE_ORDER); |
7b8ee84d8 mm: fix integer a... |
202 |
return NULL; |
7f2e9525b hugetlbfs: common... |
203 |
} |
33f2ef89f [PATCH] mm: make ... |
204 |
set_compound_page_dtor(page, free_huge_page); |
0bd0f9fb1 [PATCH] hugetlb: ... |
205 |
spin_lock(&hugetlb_lock); |
1da177e4c Linux-2.6.12-rc2 |
206 |
nr_huge_pages++; |
63b4613c3 hugetlb: fix huge... |
207 |
nr_huge_pages_node[nid]++; |
0bd0f9fb1 [PATCH] hugetlb: ... |
208 |
spin_unlock(&hugetlb_lock); |
a482289d4 [PATCH] hugepage ... |
209 |
put_page(page); /* free it into the hugepage allocator */ |
1da177e4c Linux-2.6.12-rc2 |
210 |
} |
63b4613c3 hugetlb: fix huge... |
211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 |
return page; } static int alloc_fresh_huge_page(void) { struct page *page; int start_nid; int next_nid; int ret = 0; start_nid = hugetlb_next_nid; do { page = alloc_fresh_huge_page_node(hugetlb_next_nid); if (page) ret = 1; /* * Use a helper variable to find the next node and then * copy it back to hugetlb_next_nid afterwards: * otherwise there's a window in which a racer might * pass invalid nid MAX_NUMNODES to alloc_pages_node. * But we don't need to use a spin_lock here: it really * doesn't matter if occasionally a racer chooses the * same nid as we do. Move nid forward in the mask even * if we just successfully allocated a hugepage so that * the next caller gets hugepages on the next node. */ next_nid = next_node(hugetlb_next_nid, node_online_map); if (next_nid == MAX_NUMNODES) next_nid = first_node(node_online_map); hugetlb_next_nid = next_nid; } while (!page && hugetlb_next_nid != start_nid); |
3b1163006 Subject: [PATCH] ... |
244 245 246 247 |
if (ret) count_vm_event(HTLB_BUDDY_PGALLOC); else count_vm_event(HTLB_BUDDY_PGALLOC_FAIL); |
63b4613c3 hugetlb: fix huge... |
248 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
249 |
} |
7893d1d50 hugetlb: Try to g... |
250 251 252 253 |
static struct page *alloc_buddy_huge_page(struct vm_area_struct *vma, unsigned long address) { struct page *page; |
d1c3fb1f8 hugetlb: introduc... |
254 |
unsigned int nid; |
7893d1d50 hugetlb: Try to g... |
255 |
|
d1c3fb1f8 hugetlb: introduc... |
256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 |
/* * 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); if (surplus_huge_pages >= nr_overcommit_huge_pages) { spin_unlock(&hugetlb_lock); return NULL; } else { nr_huge_pages++; surplus_huge_pages++; } spin_unlock(&hugetlb_lock); |
551883ae8 page allocator: e... |
288 289 |
page = alloc_pages(htlb_alloc_mask|__GFP_COMP| __GFP_REPEAT|__GFP_NOWARN, |
7893d1d50 hugetlb: Try to g... |
290 |
HUGETLB_PAGE_ORDER); |
d1c3fb1f8 hugetlb: introduc... |
291 292 |
spin_lock(&hugetlb_lock); |
7893d1d50 hugetlb: Try to g... |
293 |
if (page) { |
2668db911 hugetlb: correct ... |
294 295 296 297 298 299 |
/* * 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)); |
d1c3fb1f8 hugetlb: introduc... |
300 |
nid = page_to_nid(page); |
7893d1d50 hugetlb: Try to g... |
301 |
set_compound_page_dtor(page, free_huge_page); |
d1c3fb1f8 hugetlb: introduc... |
302 303 304 305 306 |
/* * We incremented the global counters already */ nr_huge_pages_node[nid]++; surplus_huge_pages_node[nid]++; |
3b1163006 Subject: [PATCH] ... |
307 |
__count_vm_event(HTLB_BUDDY_PGALLOC); |
d1c3fb1f8 hugetlb: introduc... |
308 309 310 |
} else { nr_huge_pages--; surplus_huge_pages--; |
3b1163006 Subject: [PATCH] ... |
311 |
__count_vm_event(HTLB_BUDDY_PGALLOC_FAIL); |
7893d1d50 hugetlb: Try to g... |
312 |
} |
d1c3fb1f8 hugetlb: introduc... |
313 |
spin_unlock(&hugetlb_lock); |
7893d1d50 hugetlb: Try to g... |
314 315 316 |
return page; } |
e4e574b76 hugetlb: Try to g... |
317 318 319 320 321 322 323 324 325 326 327 328 |
/* * Increase the hugetlb pool such that it can accomodate a reservation * of size 'delta'. */ static int gather_surplus_pages(int delta) { struct list_head surplus_list; struct page *page, *tmp; int ret, i; int needed, allocated; needed = (resv_huge_pages + delta) - free_huge_pages; |
ac09b3a15 hugetlb: close a ... |
329 330 |
if (needed <= 0) { resv_huge_pages += delta; |
e4e574b76 hugetlb: Try to g... |
331 |
return 0; |
ac09b3a15 hugetlb: close a ... |
332 |
} |
e4e574b76 hugetlb: Try to g... |
333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 |
allocated = 0; INIT_LIST_HEAD(&surplus_list); ret = -ENOMEM; retry: spin_unlock(&hugetlb_lock); for (i = 0; i < needed; i++) { page = alloc_buddy_huge_page(NULL, 0); if (!page) { /* * We were not able to allocate enough pages to * satisfy the entire reservation so we free what * we've allocated so far. */ spin_lock(&hugetlb_lock); needed = 0; goto free; } 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); needed = (resv_huge_pages + delta) - (free_huge_pages + allocated); 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 ... |
370 371 372 |
* 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... |
373 374 |
*/ needed += allocated; |
ac09b3a15 hugetlb: close a ... |
375 |
resv_huge_pages += delta; |
e4e574b76 hugetlb: Try to g... |
376 377 |
ret = 0; free: |
19fc3f0ac hugetlb: decrease... |
378 |
/* Free the needed pages to the hugetlb pool */ |
e4e574b76 hugetlb: Try to g... |
379 |
list_for_each_entry_safe(page, tmp, &surplus_list, lru) { |
19fc3f0ac hugetlb: decrease... |
380 381 |
if ((--needed) < 0) break; |
e4e574b76 hugetlb: Try to g... |
382 |
list_del(&page->lru); |
19fc3f0ac hugetlb: decrease... |
383 384 385 386 387 388 389 390 |
enqueue_huge_page(page); } /* Free unnecessary surplus pages to the buddy allocator */ if (!list_empty(&surplus_list)) { spin_unlock(&hugetlb_lock); list_for_each_entry_safe(page, tmp, &surplus_list, lru) { list_del(&page->lru); |
af767cbdd hugetlb: fix dyna... |
391 |
/* |
2668db911 hugetlb: correct ... |
392 393 394 |
* The page has a reference count of zero already, so * call free_huge_page directly instead of using * put_page. This must be done with hugetlb_lock |
af767cbdd hugetlb: fix dyna... |
395 396 397 |
* unlocked which is safe because free_huge_page takes * hugetlb_lock before deciding how to free the page. */ |
2668db911 hugetlb: correct ... |
398 |
free_huge_page(page); |
af767cbdd hugetlb: fix dyna... |
399 |
} |
19fc3f0ac hugetlb: decrease... |
400 |
spin_lock(&hugetlb_lock); |
e4e574b76 hugetlb: Try to g... |
401 402 403 404 405 406 407 408 409 410 |
} 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. */ |
8cde045c7 mm/hugetlb.c: mak... |
411 |
static void return_unused_surplus_pages(unsigned long unused_resv_pages) |
e4e574b76 hugetlb: Try to g... |
412 413 414 415 |
{ static int nid = -1; struct page *page; unsigned long nr_pages; |
11320d17c hugetlb: fix pote... |
416 417 418 419 420 421 422 |
/* * We want to release as many surplus pages as possible, spread * evenly across all nodes. Iterate across all nodes until we * can no longer free unreserved surplus pages. This occurs when * the nodes with surplus pages have no free pages. */ unsigned long remaining_iterations = num_online_nodes(); |
ac09b3a15 hugetlb: close a ... |
423 424 |
/* Uncommit the reservation */ resv_huge_pages -= unused_resv_pages; |
e4e574b76 hugetlb: Try to g... |
425 |
nr_pages = min(unused_resv_pages, surplus_huge_pages); |
11320d17c hugetlb: fix pote... |
426 |
while (remaining_iterations-- && nr_pages) { |
e4e574b76 hugetlb: Try to g... |
427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 |
nid = next_node(nid, node_online_map); if (nid == MAX_NUMNODES) nid = first_node(node_online_map); if (!surplus_huge_pages_node[nid]) continue; if (!list_empty(&hugepage_freelists[nid])) { page = list_entry(hugepage_freelists[nid].next, struct page, lru); list_del(&page->lru); update_and_free_page(page); free_huge_pages--; free_huge_pages_node[nid]--; surplus_huge_pages--; surplus_huge_pages_node[nid]--; nr_pages--; |
11320d17c hugetlb: fix pote... |
444 |
remaining_iterations = num_online_nodes(); |
e4e574b76 hugetlb: Try to g... |
445 446 447 |
} } } |
348ea204c hugetlb: split al... |
448 449 450 |
static struct page *alloc_huge_page_shared(struct vm_area_struct *vma, unsigned long addr) |
1da177e4c Linux-2.6.12-rc2 |
451 |
{ |
348ea204c hugetlb: split al... |
452 |
struct page *page; |
1da177e4c Linux-2.6.12-rc2 |
453 454 |
spin_lock(&hugetlb_lock); |
348e1e04b hugetlb: fix pool... |
455 |
page = dequeue_huge_page_vma(vma, addr); |
1da177e4c Linux-2.6.12-rc2 |
456 |
spin_unlock(&hugetlb_lock); |
90d8b7e61 hugetlb: enforce ... |
457 |
return page ? page : ERR_PTR(-VM_FAULT_OOM); |
348ea204c hugetlb: split al... |
458 |
} |
b45b5bd65 [PATCH] hugepage:... |
459 |
|
348ea204c hugetlb: split al... |
460 461 462 463 |
static struct page *alloc_huge_page_private(struct vm_area_struct *vma, unsigned long addr) { struct page *page = NULL; |
7893d1d50 hugetlb: Try to g... |
464 |
|
90d8b7e61 hugetlb: enforce ... |
465 466 |
if (hugetlb_get_quota(vma->vm_file->f_mapping, 1)) return ERR_PTR(-VM_FAULT_SIGBUS); |
348ea204c hugetlb: split al... |
467 468 |
spin_lock(&hugetlb_lock); if (free_huge_pages > resv_huge_pages) |
348e1e04b hugetlb: fix pool... |
469 |
page = dequeue_huge_page_vma(vma, addr); |
348ea204c hugetlb: split al... |
470 |
spin_unlock(&hugetlb_lock); |
68842c9b9 hugetlbfs: fix qu... |
471 |
if (!page) { |
7893d1d50 hugetlb: Try to g... |
472 |
page = alloc_buddy_huge_page(vma, addr); |
68842c9b9 hugetlbfs: fix qu... |
473 474 475 476 477 478 |
if (!page) { hugetlb_put_quota(vma->vm_file->f_mapping, 1); return ERR_PTR(-VM_FAULT_OOM); } } return page; |
348ea204c hugetlb: split al... |
479 480 481 482 483 484 |
} static struct page *alloc_huge_page(struct vm_area_struct *vma, unsigned long addr) { struct page *page; |
2fc39cec6 hugetlb: debit qu... |
485 |
struct address_space *mapping = vma->vm_file->f_mapping; |
348ea204c hugetlb: split al... |
486 487 488 489 |
if (vma->vm_flags & VM_MAYSHARE) page = alloc_huge_page_shared(vma, addr); else page = alloc_huge_page_private(vma, addr); |
90d8b7e61 hugetlb: enforce ... |
490 491 |
if (!IS_ERR(page)) { |
348ea204c hugetlb: split al... |
492 |
set_page_refcounted(page); |
2fc39cec6 hugetlb: debit qu... |
493 |
set_page_private(page, (unsigned long) mapping); |
90d8b7e61 hugetlb: enforce ... |
494 495 |
} return page; |
b45b5bd65 [PATCH] hugepage:... |
496 |
} |
1da177e4c Linux-2.6.12-rc2 |
497 498 499 |
static int __init hugetlb_init(void) { unsigned long i; |
1da177e4c Linux-2.6.12-rc2 |
500 |
|
3c726f8de [PATCH] ppc64: su... |
501 502 |
if (HPAGE_SHIFT == 0) return 0; |
1da177e4c Linux-2.6.12-rc2 |
503 504 |
for (i = 0; i < MAX_NUMNODES; ++i) INIT_LIST_HEAD(&hugepage_freelists[i]); |
63b4613c3 hugetlb: fix huge... |
505 |
hugetlb_next_nid = first_node(node_online_map); |
1da177e4c Linux-2.6.12-rc2 |
506 |
for (i = 0; i < max_huge_pages; ++i) { |
a482289d4 [PATCH] hugepage ... |
507 |
if (!alloc_fresh_huge_page()) |
1da177e4c Linux-2.6.12-rc2 |
508 |
break; |
1da177e4c Linux-2.6.12-rc2 |
509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 |
} max_huge_pages = free_huge_pages = nr_huge_pages = i; printk("Total HugeTLB memory allocated, %ld ", free_huge_pages); return 0; } module_init(hugetlb_init); static int __init hugetlb_setup(char *s) { if (sscanf(s, "%lu", &max_huge_pages) <= 0) max_huge_pages = 0; return 1; } __setup("hugepages=", hugetlb_setup); |
8a6301127 pretend cpuset ha... |
524 525 526 527 528 529 530 531 532 533 |
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; } |
1da177e4c Linux-2.6.12-rc2 |
534 |
#ifdef CONFIG_SYSCTL |
1da177e4c Linux-2.6.12-rc2 |
535 536 537 |
#ifdef CONFIG_HIGHMEM static void try_to_free_low(unsigned long count) { |
4415cc8df [PATCH] Hugepages... |
538 |
int i; |
1da177e4c Linux-2.6.12-rc2 |
539 540 541 |
for (i = 0; i < MAX_NUMNODES; ++i) { struct page *page, *next; list_for_each_entry_safe(page, next, &hugepage_freelists[i], lru) { |
6b0c880df hugetlb: fix pool... |
542 543 |
if (count >= nr_huge_pages) return; |
1da177e4c Linux-2.6.12-rc2 |
544 545 546 547 |
if (PageHighMem(page)) continue; list_del(&page->lru); update_and_free_page(page); |
1da177e4c Linux-2.6.12-rc2 |
548 |
free_huge_pages--; |
4415cc8df [PATCH] Hugepages... |
549 |
free_huge_pages_node[page_to_nid(page)]--; |
1da177e4c Linux-2.6.12-rc2 |
550 551 552 553 554 555 556 557 |
} } } #else static inline void try_to_free_low(unsigned long count) { } #endif |
7893d1d50 hugetlb: Try to g... |
558 |
#define persistent_huge_pages (nr_huge_pages - surplus_huge_pages) |
1da177e4c Linux-2.6.12-rc2 |
559 560 |
static unsigned long set_max_huge_pages(unsigned long count) { |
7893d1d50 hugetlb: Try to g... |
561 |
unsigned long min_count, ret; |
1da177e4c Linux-2.6.12-rc2 |
562 |
|
7893d1d50 hugetlb: Try to g... |
563 564 565 566 |
/* * 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... |
567 568 569 570 571 572 |
* * 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... |
573 |
*/ |
1da177e4c Linux-2.6.12-rc2 |
574 |
spin_lock(&hugetlb_lock); |
7893d1d50 hugetlb: Try to g... |
575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 |
while (surplus_huge_pages && count > persistent_huge_pages) { if (!adjust_pool_surplus(-1)) break; } while (count > persistent_huge_pages) { int ret; /* * 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); ret = alloc_fresh_huge_page(); spin_lock(&hugetlb_lock); if (!ret) goto out; } |
7893d1d50 hugetlb: Try to g... |
594 595 596 597 598 599 600 |
/* * 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... |
601 602 603 604 605 606 607 608 |
* * 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... |
609 |
*/ |
6b0c880df hugetlb: fix pool... |
610 611 |
min_count = resv_huge_pages + nr_huge_pages - free_huge_pages; min_count = max(count, min_count); |
7893d1d50 hugetlb: Try to g... |
612 613 |
try_to_free_low(min_count); while (min_count < persistent_huge_pages) { |
348e1e04b hugetlb: fix pool... |
614 |
struct page *page = dequeue_huge_page(); |
1da177e4c Linux-2.6.12-rc2 |
615 616 617 618 |
if (!page) break; update_and_free_page(page); } |
7893d1d50 hugetlb: Try to g... |
619 620 621 622 623 624 |
while (count < persistent_huge_pages) { if (!adjust_pool_surplus(1)) break; } out: ret = persistent_huge_pages; |
1da177e4c Linux-2.6.12-rc2 |
625 |
spin_unlock(&hugetlb_lock); |
7893d1d50 hugetlb: Try to g... |
626 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
627 628 629 630 631 632 633 634 635 636 |
} int hugetlb_sysctl_handler(struct ctl_table *table, int write, struct file *file, void __user *buffer, size_t *length, loff_t *ppos) { proc_doulongvec_minmax(table, write, file, buffer, length, ppos); max_huge_pages = set_max_huge_pages(max_huge_pages); return 0; } |
396faf030 Allow huge page a... |
637 638 639 640 641 642 643 644 645 646 647 648 |
int hugetlb_treat_movable_handler(struct ctl_table *table, int write, struct file *file, void __user *buffer, size_t *length, loff_t *ppos) { proc_dointvec(table, write, file, buffer, length, ppos); if (hugepages_treat_as_movable) htlb_alloc_mask = GFP_HIGHUSER_MOVABLE; else htlb_alloc_mask = GFP_HIGHUSER; return 0; } |
a3d0c6aa1 hugetlb: add lock... |
649 650 651 652 |
int hugetlb_overcommit_handler(struct ctl_table *table, int write, struct file *file, void __user *buffer, size_t *length, loff_t *ppos) { |
a3d0c6aa1 hugetlb: add lock... |
653 |
proc_doulongvec_minmax(table, write, file, buffer, length, ppos); |
064d9efe9 hugetlb: fix over... |
654 655 |
spin_lock(&hugetlb_lock); nr_overcommit_huge_pages = sysctl_overcommit_huge_pages; |
a3d0c6aa1 hugetlb: add lock... |
656 657 658 |
spin_unlock(&hugetlb_lock); return 0; } |
1da177e4c Linux-2.6.12-rc2 |
659 660 661 662 663 664 665 666 667 |
#endif /* CONFIG_SYSCTL */ int hugetlb_report_meminfo(char *buf) { return sprintf(buf, "HugePages_Total: %5lu " "HugePages_Free: %5lu " |
a43a8c39b [PATCH] tightenin... |
668 669 |
"HugePages_Rsvd: %5lu " |
7893d1d50 hugetlb: Try to g... |
670 671 |
"HugePages_Surp: %5lu " |
1da177e4c Linux-2.6.12-rc2 |
672 673 674 675 |
"Hugepagesize: %5lu kB ", nr_huge_pages, free_huge_pages, |
a43a8c39b [PATCH] tightenin... |
676 |
resv_huge_pages, |
7893d1d50 hugetlb: Try to g... |
677 |
surplus_huge_pages, |
1da177e4c Linux-2.6.12-rc2 |
678 679 680 681 682 683 684 685 |
HPAGE_SIZE/1024); } int hugetlb_report_node_meminfo(int nid, char *buf) { return sprintf(buf, "Node %d HugePages_Total: %5u " |
a1de09195 hugetlb: indicate... |
686 687 688 689 |
"Node %d HugePages_Free: %5u " "Node %d HugePages_Surp: %5u ", |
1da177e4c Linux-2.6.12-rc2 |
690 |
nid, nr_huge_pages_node[nid], |
a1de09195 hugetlb: indicate... |
691 692 |
nid, free_huge_pages_node[nid], nid, surplus_huge_pages_node[nid]); |
1da177e4c Linux-2.6.12-rc2 |
693 |
} |
1da177e4c Linux-2.6.12-rc2 |
694 695 696 697 698 |
/* Return the number pages of memory we physically have, in PAGE_SIZE units. */ unsigned long hugetlb_total_pages(void) { return nr_huge_pages * (HPAGE_SIZE / PAGE_SIZE); } |
1da177e4c Linux-2.6.12-rc2 |
699 700 701 702 703 704 705 |
/* * 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... |
706 |
static int hugetlb_vm_op_fault(struct vm_area_struct *vma, struct vm_fault *vmf) |
1da177e4c Linux-2.6.12-rc2 |
707 708 |
{ BUG(); |
d0217ac04 mm: fault feedbac... |
709 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
710 711 712 |
} struct vm_operations_struct hugetlb_vm_ops = { |
d0217ac04 mm: fault feedbac... |
713 |
.fault = hugetlb_vm_op_fault, |
1da177e4c Linux-2.6.12-rc2 |
714 |
}; |
1e8f889b1 [PATCH] Hugetlb: ... |
715 716 |
static pte_t make_huge_pte(struct vm_area_struct *vma, struct page *page, int writable) |
63551ae0f [PATCH] Hugepage ... |
717 718 |
{ pte_t entry; |
1e8f889b1 [PATCH] Hugetlb: ... |
719 |
if (writable) { |
63551ae0f [PATCH] Hugepage ... |
720 721 722 |
entry = pte_mkwrite(pte_mkdirty(mk_pte(page, vma->vm_page_prot))); } else { |
7f2e9525b hugetlbfs: common... |
723 |
entry = huge_pte_wrprotect(mk_pte(page, vma->vm_page_prot)); |
63551ae0f [PATCH] Hugepage ... |
724 725 726 727 728 729 |
} entry = pte_mkyoung(entry); entry = pte_mkhuge(entry); return entry; } |
1e8f889b1 [PATCH] Hugetlb: ... |
730 731 732 733 |
static void set_huge_ptep_writable(struct vm_area_struct *vma, unsigned long address, pte_t *ptep) { pte_t entry; |
7f2e9525b hugetlbfs: common... |
734 735 |
entry = pte_mkwrite(pte_mkdirty(huge_ptep_get(ptep))); if (huge_ptep_set_access_flags(vma, address, ptep, entry, 1)) { |
8dab5241d Rework ptep_set_a... |
736 |
update_mmu_cache(vma, address, entry); |
8dab5241d Rework ptep_set_a... |
737 |
} |
1e8f889b1 [PATCH] Hugetlb: ... |
738 |
} |
63551ae0f [PATCH] Hugepage ... |
739 740 741 742 743 |
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... |
744 |
unsigned long addr; |
1e8f889b1 [PATCH] Hugetlb: ... |
745 746 747 |
int cow; cow = (vma->vm_flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE; |
63551ae0f [PATCH] Hugepage ... |
748 |
|
1c59827d1 [PATCH] mm: huget... |
749 |
for (addr = vma->vm_start; addr < vma->vm_end; addr += HPAGE_SIZE) { |
c74df32c7 [PATCH] mm: ptd_a... |
750 751 752 |
src_pte = huge_pte_offset(src, addr); if (!src_pte) continue; |
63551ae0f [PATCH] Hugepage ... |
753 754 755 |
dst_pte = huge_pte_alloc(dst, addr); if (!dst_pte) goto nomem; |
c5c99429f fix hugepages lea... |
756 757 758 759 |
/* If the pagetables are shared don't copy or take references */ if (dst_pte == src_pte) continue; |
c74df32c7 [PATCH] mm: ptd_a... |
760 |
spin_lock(&dst->page_table_lock); |
464787581 hugetlb: fix lock... |
761 |
spin_lock_nested(&src->page_table_lock, SINGLE_DEPTH_NESTING); |
7f2e9525b hugetlbfs: common... |
762 |
if (!huge_pte_none(huge_ptep_get(src_pte))) { |
1e8f889b1 [PATCH] Hugetlb: ... |
763 |
if (cow) |
7f2e9525b hugetlbfs: common... |
764 765 |
huge_ptep_set_wrprotect(src, addr, src_pte); entry = huge_ptep_get(src_pte); |
1c59827d1 [PATCH] mm: huget... |
766 767 |
ptepage = pte_page(entry); get_page(ptepage); |
1c59827d1 [PATCH] mm: huget... |
768 769 770 |
set_huge_pte_at(dst, addr, dst_pte, entry); } spin_unlock(&src->page_table_lock); |
c74df32c7 [PATCH] mm: ptd_a... |
771 |
spin_unlock(&dst->page_table_lock); |
63551ae0f [PATCH] Hugepage ... |
772 773 774 775 776 777 |
} return 0; nomem: return -ENOMEM; } |
502717f4e [PATCH] hugetlb: ... |
778 779 |
void __unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start, unsigned long end) |
63551ae0f [PATCH] Hugepage ... |
780 781 782 |
{ struct mm_struct *mm = vma->vm_mm; unsigned long address; |
c7546f8f0 [PATCH] Fix hugep... |
783 |
pte_t *ptep; |
63551ae0f [PATCH] Hugepage ... |
784 785 |
pte_t pte; struct page *page; |
fe1668ae5 [PATCH] enforce p... |
786 |
struct page *tmp; |
c0a499c2c [PATCH] __unmap_h... |
787 788 789 790 791 |
/* * 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... |
792 |
LIST_HEAD(page_list); |
63551ae0f [PATCH] Hugepage ... |
793 794 795 796 |
WARN_ON(!is_vm_hugetlb_page(vma)); BUG_ON(start & ~HPAGE_MASK); BUG_ON(end & ~HPAGE_MASK); |
508034a32 [PATCH] mm: unmap... |
797 |
spin_lock(&mm->page_table_lock); |
63551ae0f [PATCH] Hugepage ... |
798 |
for (address = start; address < end; address += HPAGE_SIZE) { |
c7546f8f0 [PATCH] Fix hugep... |
799 |
ptep = huge_pte_offset(mm, address); |
4c8872659 [PATCH] hugetlb: ... |
800 |
if (!ptep) |
c7546f8f0 [PATCH] Fix hugep... |
801 |
continue; |
39dde65c9 [PATCH] shared pa... |
802 803 |
if (huge_pmd_unshare(mm, &address, ptep)) continue; |
c7546f8f0 [PATCH] Fix hugep... |
804 |
pte = huge_ptep_get_and_clear(mm, address, ptep); |
7f2e9525b hugetlbfs: common... |
805 |
if (huge_pte_none(pte)) |
63551ae0f [PATCH] Hugepage ... |
806 |
continue; |
c7546f8f0 [PATCH] Fix hugep... |
807 |
|
63551ae0f [PATCH] Hugepage ... |
808 |
page = pte_page(pte); |
6649a3863 [PATCH] hugetlb: ... |
809 810 |
if (pte_dirty(pte)) set_page_dirty(page); |
fe1668ae5 [PATCH] enforce p... |
811 |
list_add(&page->lru, &page_list); |
63551ae0f [PATCH] Hugepage ... |
812 |
} |
1da177e4c Linux-2.6.12-rc2 |
813 |
spin_unlock(&mm->page_table_lock); |
508034a32 [PATCH] mm: unmap... |
814 |
flush_tlb_range(vma, start, end); |
fe1668ae5 [PATCH] enforce p... |
815 816 817 818 |
list_for_each_entry_safe(page, tmp, &page_list, lru) { list_del(&page->lru); put_page(page); } |
1da177e4c Linux-2.6.12-rc2 |
819 |
} |
63551ae0f [PATCH] Hugepage ... |
820 |
|
502717f4e [PATCH] hugetlb: ... |
821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 |
void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start, unsigned long end) { /* * It is undesirable to test vma->vm_file as it should be non-null * for valid hugetlb area. However, vm_file will be NULL in the error * cleanup path of do_mmap_pgoff. When hugetlbfs ->mmap method fails, * do_mmap_pgoff() nullifies vma->vm_file before calling this function * to clean up. Since no pte has actually been setup, it is safe to * do nothing in this case. */ if (vma->vm_file) { spin_lock(&vma->vm_file->f_mapping->i_mmap_lock); __unmap_hugepage_range(vma, start, end); spin_unlock(&vma->vm_file->f_mapping->i_mmap_lock); } } |
1e8f889b1 [PATCH] Hugetlb: ... |
838 839 840 841 |
static int hugetlb_cow(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long address, pte_t *ptep, pte_t pte) { struct page *old_page, *new_page; |
79ac6ba40 [PATCH] hugepage:... |
842 |
int avoidcopy; |
1e8f889b1 [PATCH] Hugetlb: ... |
843 844 845 846 847 848 849 850 |
old_page = pte_page(pte); /* If no-one else is actually using this page, avoid the copy * and just make the page writable */ avoidcopy = (page_count(old_page) == 1); if (avoidcopy) { set_huge_ptep_writable(vma, address, ptep); |
83c54070e mm: fault feedbac... |
851 |
return 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
852 853 854 |
} page_cache_get(old_page); |
5da7ca860 [PATCH] Add NUMA ... |
855 |
new_page = alloc_huge_page(vma, address); |
1e8f889b1 [PATCH] Hugetlb: ... |
856 |
|
2fc39cec6 hugetlb: debit qu... |
857 |
if (IS_ERR(new_page)) { |
1e8f889b1 [PATCH] Hugetlb: ... |
858 |
page_cache_release(old_page); |
2fc39cec6 hugetlb: debit qu... |
859 |
return -PTR_ERR(new_page); |
1e8f889b1 [PATCH] Hugetlb: ... |
860 861 862 |
} spin_unlock(&mm->page_table_lock); |
9de455b20 [PATCH] Pass vma ... |
863 |
copy_huge_page(new_page, old_page, address, vma); |
0ed361dec mm: fix PageUptod... |
864 |
__SetPageUptodate(new_page); |
1e8f889b1 [PATCH] Hugetlb: ... |
865 866 867 |
spin_lock(&mm->page_table_lock); ptep = huge_pte_offset(mm, address & HPAGE_MASK); |
7f2e9525b hugetlbfs: common... |
868 |
if (likely(pte_same(huge_ptep_get(ptep), pte))) { |
1e8f889b1 [PATCH] Hugetlb: ... |
869 |
/* Break COW */ |
8fe627ec5 hugetlbfs: add mi... |
870 |
huge_ptep_clear_flush(vma, address, ptep); |
1e8f889b1 [PATCH] Hugetlb: ... |
871 872 873 874 875 876 877 |
set_huge_pte_at(mm, address, ptep, make_huge_pte(vma, new_page, 1)); /* Make the old page be freed below */ new_page = old_page; } page_cache_release(new_page); page_cache_release(old_page); |
83c54070e mm: fault feedbac... |
878 |
return 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
879 |
} |
a1ed3dda0 MM: Make needless... |
880 |
static int hugetlb_no_page(struct mm_struct *mm, struct vm_area_struct *vma, |
1e8f889b1 [PATCH] Hugetlb: ... |
881 |
unsigned long address, pte_t *ptep, int write_access) |
ac9b9c667 [PATCH] Fix handl... |
882 883 |
{ int ret = VM_FAULT_SIGBUS; |
4c8872659 [PATCH] hugetlb: ... |
884 885 |
unsigned long idx; unsigned long size; |
4c8872659 [PATCH] hugetlb: ... |
886 887 |
struct page *page; struct address_space *mapping; |
1e8f889b1 [PATCH] Hugetlb: ... |
888 |
pte_t new_pte; |
4c8872659 [PATCH] hugetlb: ... |
889 |
|
4c8872659 [PATCH] hugetlb: ... |
890 891 892 893 894 895 896 897 |
mapping = vma->vm_file->f_mapping; idx = ((address - vma->vm_start) >> HPAGE_SHIFT) + (vma->vm_pgoff >> (HPAGE_SHIFT - PAGE_SHIFT)); /* * Use page lock to guard against racing truncation * before we get page_table_lock. */ |
6bda666a0 [PATCH] hugepages... |
898 899 900 |
retry: page = find_lock_page(mapping, idx); if (!page) { |
ebed4bfc8 [PATCH] hugetlb: ... |
901 902 903 |
size = i_size_read(mapping->host) >> HPAGE_SHIFT; if (idx >= size) goto out; |
6bda666a0 [PATCH] hugepages... |
904 |
page = alloc_huge_page(vma, address); |
2fc39cec6 hugetlb: debit qu... |
905 906 |
if (IS_ERR(page)) { ret = -PTR_ERR(page); |
6bda666a0 [PATCH] hugepages... |
907 908 |
goto out; } |
79ac6ba40 [PATCH] hugepage:... |
909 |
clear_huge_page(page, address); |
0ed361dec mm: fix PageUptod... |
910 |
__SetPageUptodate(page); |
ac9b9c667 [PATCH] Fix handl... |
911 |
|
6bda666a0 [PATCH] hugepages... |
912 913 |
if (vma->vm_flags & VM_SHARED) { int err; |
45c682a68 hugetlb: fix i_bl... |
914 |
struct inode *inode = mapping->host; |
6bda666a0 [PATCH] hugepages... |
915 916 917 918 |
err = add_to_page_cache(page, mapping, idx, GFP_KERNEL); if (err) { put_page(page); |
6bda666a0 [PATCH] hugepages... |
919 920 921 922 |
if (err == -EEXIST) goto retry; goto out; } |
45c682a68 hugetlb: fix i_bl... |
923 924 925 926 |
spin_lock(&inode->i_lock); inode->i_blocks += BLOCKS_PER_HUGEPAGE; spin_unlock(&inode->i_lock); |
6bda666a0 [PATCH] hugepages... |
927 928 929 |
} else lock_page(page); } |
1e8f889b1 [PATCH] Hugetlb: ... |
930 |
|
ac9b9c667 [PATCH] Fix handl... |
931 |
spin_lock(&mm->page_table_lock); |
4c8872659 [PATCH] hugetlb: ... |
932 933 934 |
size = i_size_read(mapping->host) >> HPAGE_SHIFT; if (idx >= size) goto backout; |
83c54070e mm: fault feedbac... |
935 |
ret = 0; |
7f2e9525b hugetlbfs: common... |
936 |
if (!huge_pte_none(huge_ptep_get(ptep))) |
4c8872659 [PATCH] hugetlb: ... |
937 |
goto backout; |
1e8f889b1 [PATCH] Hugetlb: ... |
938 939 940 941 942 943 944 945 |
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); if (write_access && !(vma->vm_flags & VM_SHARED)) { /* Optimization, do the COW without a second fault */ ret = hugetlb_cow(mm, vma, address, ptep, new_pte); } |
ac9b9c667 [PATCH] Fix handl... |
946 |
spin_unlock(&mm->page_table_lock); |
4c8872659 [PATCH] hugetlb: ... |
947 948 |
unlock_page(page); out: |
ac9b9c667 [PATCH] Fix handl... |
949 |
return ret; |
4c8872659 [PATCH] hugetlb: ... |
950 951 952 |
backout: spin_unlock(&mm->page_table_lock); |
4c8872659 [PATCH] hugetlb: ... |
953 954 955 |
unlock_page(page); put_page(page); goto out; |
ac9b9c667 [PATCH] Fix handl... |
956 |
} |
86e5216f8 [PATCH] Hugetlb: ... |
957 958 959 960 961 |
int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long address, int write_access) { pte_t *ptep; pte_t entry; |
1e8f889b1 [PATCH] Hugetlb: ... |
962 |
int ret; |
3935baa9b [PATCH] hugepage:... |
963 |
static DEFINE_MUTEX(hugetlb_instantiation_mutex); |
86e5216f8 [PATCH] Hugetlb: ... |
964 965 966 967 |
ptep = huge_pte_alloc(mm, address); if (!ptep) return VM_FAULT_OOM; |
3935baa9b [PATCH] hugepage:... |
968 969 970 971 972 973 |
/* * 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... |
974 975 |
entry = huge_ptep_get(ptep); if (huge_pte_none(entry)) { |
3935baa9b [PATCH] hugepage:... |
976 977 978 979 |
ret = hugetlb_no_page(mm, vma, address, ptep, write_access); mutex_unlock(&hugetlb_instantiation_mutex); return ret; } |
86e5216f8 [PATCH] Hugetlb: ... |
980 |
|
83c54070e mm: fault feedbac... |
981 |
ret = 0; |
1e8f889b1 [PATCH] Hugetlb: ... |
982 983 984 |
spin_lock(&mm->page_table_lock); /* Check for a racing update before calling hugetlb_cow */ |
7f2e9525b hugetlbfs: common... |
985 |
if (likely(pte_same(entry, huge_ptep_get(ptep)))) |
1e8f889b1 [PATCH] Hugetlb: ... |
986 987 988 |
if (write_access && !pte_write(entry)) ret = hugetlb_cow(mm, vma, address, ptep, entry); spin_unlock(&mm->page_table_lock); |
3935baa9b [PATCH] hugepage:... |
989 |
mutex_unlock(&hugetlb_instantiation_mutex); |
1e8f889b1 [PATCH] Hugetlb: ... |
990 991 |
return ret; |
86e5216f8 [PATCH] Hugetlb: ... |
992 |
} |
63551ae0f [PATCH] Hugepage ... |
993 994 |
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... |
995 996 |
unsigned long *position, int *length, int i, int write) |
63551ae0f [PATCH] Hugepage ... |
997 |
{ |
d5d4b0aa4 [PATCH] optimize ... |
998 999 |
unsigned long pfn_offset; unsigned long vaddr = *position; |
63551ae0f [PATCH] Hugepage ... |
1000 |
int remainder = *length; |
1c59827d1 [PATCH] mm: huget... |
1001 |
spin_lock(&mm->page_table_lock); |
63551ae0f [PATCH] Hugepage ... |
1002 |
while (vaddr < vma->vm_end && remainder) { |
4c8872659 [PATCH] hugetlb: ... |
1003 1004 |
pte_t *pte; struct page *page; |
63551ae0f [PATCH] Hugepage ... |
1005 |
|
4c8872659 [PATCH] hugetlb: ... |
1006 1007 1008 1009 1010 1011 |
/* * Some archs (sparc64, sh*) have multiple pte_ts to * each hugepage. We have to make * sure we get the * first, for the page indexing below to work. */ pte = huge_pte_offset(mm, vaddr & HPAGE_MASK); |
63551ae0f [PATCH] Hugepage ... |
1012 |
|
7f2e9525b hugetlbfs: common... |
1013 1014 |
if (!pte || huge_pte_none(huge_ptep_get(pte)) || (write && !pte_write(huge_ptep_get(pte)))) { |
4c8872659 [PATCH] hugetlb: ... |
1015 |
int ret; |
63551ae0f [PATCH] Hugepage ... |
1016 |
|
4c8872659 [PATCH] hugetlb: ... |
1017 |
spin_unlock(&mm->page_table_lock); |
5b23dbe81 hugetlb: follow_h... |
1018 |
ret = hugetlb_fault(mm, vma, vaddr, write); |
4c8872659 [PATCH] hugetlb: ... |
1019 |
spin_lock(&mm->page_table_lock); |
a89182c76 Fix VM_FAULT flag... |
1020 |
if (!(ret & VM_FAULT_ERROR)) |
4c8872659 [PATCH] hugetlb: ... |
1021 |
continue; |
63551ae0f [PATCH] Hugepage ... |
1022 |
|
4c8872659 [PATCH] hugetlb: ... |
1023 1024 1025 1026 1027 |
remainder = 0; if (!i) i = -EFAULT; break; } |
d5d4b0aa4 [PATCH] optimize ... |
1028 |
pfn_offset = (vaddr & ~HPAGE_MASK) >> PAGE_SHIFT; |
7f2e9525b hugetlbfs: common... |
1029 |
page = pte_page(huge_ptep_get(pte)); |
d5d4b0aa4 [PATCH] optimize ... |
1030 |
same_page: |
d6692183a [PATCH] fix extra... |
1031 1032 |
if (pages) { get_page(page); |
d5d4b0aa4 [PATCH] optimize ... |
1033 |
pages[i] = page + pfn_offset; |
d6692183a [PATCH] fix extra... |
1034 |
} |
63551ae0f [PATCH] Hugepage ... |
1035 1036 1037 1038 1039 |
if (vmas) vmas[i] = vma; vaddr += PAGE_SIZE; |
d5d4b0aa4 [PATCH] optimize ... |
1040 |
++pfn_offset; |
63551ae0f [PATCH] Hugepage ... |
1041 1042 |
--remainder; ++i; |
d5d4b0aa4 [PATCH] optimize ... |
1043 1044 1045 1046 1047 1048 1049 1050 |
if (vaddr < vma->vm_end && remainder && pfn_offset < HPAGE_SIZE/PAGE_SIZE) { /* * We use pfn_offset to avoid touching the pageframes * of this compound page. */ goto same_page; } |
63551ae0f [PATCH] Hugepage ... |
1051 |
} |
1c59827d1 [PATCH] mm: huget... |
1052 |
spin_unlock(&mm->page_table_lock); |
63551ae0f [PATCH] Hugepage ... |
1053 1054 1055 1056 1057 |
*length = remainder; *position = vaddr; return i; } |
8f860591f [PATCH] Enable mp... |
1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 |
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; BUG_ON(address >= end); flush_cache_range(vma, address, end); |
39dde65c9 [PATCH] shared pa... |
1069 |
spin_lock(&vma->vm_file->f_mapping->i_mmap_lock); |
8f860591f [PATCH] Enable mp... |
1070 1071 1072 1073 1074 |
spin_lock(&mm->page_table_lock); for (; address < end; address += HPAGE_SIZE) { ptep = huge_pte_offset(mm, address); if (!ptep) continue; |
39dde65c9 [PATCH] shared pa... |
1075 1076 |
if (huge_pmd_unshare(mm, &address, ptep)) continue; |
7f2e9525b hugetlbfs: common... |
1077 |
if (!huge_pte_none(huge_ptep_get(ptep))) { |
8f860591f [PATCH] Enable mp... |
1078 1079 1080 |
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... |
1081 1082 1083 |
} } spin_unlock(&mm->page_table_lock); |
39dde65c9 [PATCH] shared pa... |
1084 |
spin_unlock(&vma->vm_file->f_mapping->i_mmap_lock); |
8f860591f [PATCH] Enable mp... |
1085 1086 1087 |
flush_tlb_range(vma, start, end); } |
a43a8c39b [PATCH] tightenin... |
1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 |
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 |
183ff22bb spelling fixes: mm/ |
1142 |
* size such that we can guarantee to record the reservation. */ |
a43a8c39b [PATCH] tightenin... |
1143 1144 |
if (&rg->link == head || t < rg->from) { nrg = kmalloc(sizeof(*nrg), GFP_KERNEL); |
c80544dc0 sparse pointer us... |
1145 |
if (!nrg) |
a43a8c39b [PATCH] tightenin... |
1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 |
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; } static int hugetlb_acct_memory(long delta) { int ret = -ENOMEM; spin_lock(&hugetlb_lock); |
8a6301127 pretend cpuset ha... |
1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 |
/* * 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. */ |
e4e574b76 hugetlb: Try to g... |
1231 1232 1233 |
if (delta > 0) { if (gather_surplus_pages(delta) < 0) goto out; |
ac09b3a15 hugetlb: close a ... |
1234 1235 |
if (delta > cpuset_mems_nr(free_huge_pages_node)) { return_unused_surplus_pages(delta); |
e4e574b76 hugetlb: Try to g... |
1236 |
goto out; |
ac09b3a15 hugetlb: close a ... |
1237 |
} |
e4e574b76 hugetlb: Try to g... |
1238 1239 1240 |
} ret = 0; |
e4e574b76 hugetlb: Try to g... |
1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 |
if (delta < 0) return_unused_surplus_pages((unsigned long) -delta); out: spin_unlock(&hugetlb_lock); return ret; } int hugetlb_reserve_pages(struct inode *inode, long from, long to) { long ret, chg; chg = region_chg(&inode->i_mapping->private_list, from, to); if (chg < 0) return chg; |
8a6301127 pretend cpuset ha... |
1256 |
|
90d8b7e61 hugetlb: enforce ... |
1257 1258 |
if (hugetlb_get_quota(inode->i_mapping, chg)) return -ENOSPC; |
a43a8c39b [PATCH] tightenin... |
1259 |
ret = hugetlb_acct_memory(chg); |
68842c9b9 hugetlbfs: fix qu... |
1260 1261 |
if (ret < 0) { hugetlb_put_quota(inode->i_mapping, chg); |
a43a8c39b [PATCH] tightenin... |
1262 |
return ret; |
68842c9b9 hugetlbfs: fix qu... |
1263 |
} |
a43a8c39b [PATCH] tightenin... |
1264 1265 1266 1267 1268 1269 1270 |
region_add(&inode->i_mapping->private_list, from, to); return 0; } void hugetlb_unreserve_pages(struct inode *inode, long offset, long freed) { long chg = region_truncate(&inode->i_mapping->private_list, offset); |
45c682a68 hugetlb: fix i_bl... |
1271 1272 1273 1274 |
spin_lock(&inode->i_lock); inode->i_blocks -= BLOCKS_PER_HUGEPAGE * freed; spin_unlock(&inode->i_lock); |
90d8b7e61 hugetlb: enforce ... |
1275 1276 |
hugetlb_put_quota(inode->i_mapping, (chg - freed)); hugetlb_acct_memory(-(chg - freed)); |
a43a8c39b [PATCH] tightenin... |
1277 |
} |