Blame view
mm/madvise.c
32.4 KB
b24413180 License cleanup: ... |
1 |
// SPDX-License-Identifier: GPL-2.0 |
1da177e4c Linux-2.6.12-rc2 |
2 3 4 5 6 7 8 9 10 11 |
/* * linux/mm/madvise.c * * Copyright (C) 1999 Linus Torvalds * Copyright (C) 2002 Christoph Hellwig */ #include <linux/mman.h> #include <linux/pagemap.h> #include <linux/syscalls.h> |
05b743847 [PATCH] madvise: ... |
12 |
#include <linux/mempolicy.h> |
afcf938ee HWPOISON: Add a m... |
13 |
#include <linux/page-isolation.h> |
9c276cc65 mm: introduce MAD... |
14 |
#include <linux/page_idle.h> |
05ce77249 userfaultfd: non-... |
15 |
#include <linux/userfaultfd_k.h> |
1da177e4c Linux-2.6.12-rc2 |
16 |
#include <linux/hugetlb.h> |
3f31d0757 mm/fs: route MADV... |
17 |
#include <linux/falloc.h> |
692fe6243 mm: Handle MADV_W... |
18 |
#include <linux/fadvise.h> |
e8edc6e03 Detach sched.h fr... |
19 |
#include <linux/sched.h> |
ecb8ac8b1 mm/madvise: intro... |
20 21 |
#include <linux/sched/mm.h> #include <linux/uio.h> |
f8af4da3b ksm: the mm inter... |
22 |
#include <linux/ksm.h> |
3f31d0757 mm/fs: route MADV... |
23 |
#include <linux/fs.h> |
9ab4233dd mm: Hold a file r... |
24 |
#include <linux/file.h> |
1998cc048 mm: make madvise(... |
25 |
#include <linux/blkdev.h> |
66114cad6 writeback: separa... |
26 |
#include <linux/backing-dev.h> |
a520110e4 mm: split out a n... |
27 |
#include <linux/pagewalk.h> |
1998cc048 mm: make madvise(... |
28 29 |
#include <linux/swap.h> #include <linux/swapops.h> |
3a4f8a0b3 mm: remove shmem_... |
30 |
#include <linux/shmem_fs.h> |
854e9ed09 mm: support madvi... |
31 32 33 |
#include <linux/mmu_notifier.h> #include <asm/tlb.h> |
1da177e4c Linux-2.6.12-rc2 |
34 |
|
235190738 oom-reaper: use m... |
35 |
#include "internal.h" |
d616d5126 mm: factor out co... |
36 37 38 39 |
struct madvise_walk_private { struct mmu_gather *tlb; bool pageout; }; |
1da177e4c Linux-2.6.12-rc2 |
40 |
/* |
0a27a14a6 mm: madvise avoid... |
41 |
* Any behaviour which results in changes to the vma->vm_flags needs to |
c1e8d7c6a mmap locking API:... |
42 |
* take mmap_lock for writing. Others, which simply traverse vmas, need |
0a27a14a6 mm: madvise avoid... |
43 44 45 46 47 48 49 50 |
* to only take it for reading. */ static int madvise_need_mmap_write(int behavior) { switch (behavior) { case MADV_REMOVE: case MADV_WILLNEED: case MADV_DONTNEED: |
9c276cc65 mm: introduce MAD... |
51 |
case MADV_COLD: |
1a4e58cce mm: introduce MAD... |
52 |
case MADV_PAGEOUT: |
854e9ed09 mm: support madvi... |
53 |
case MADV_FREE: |
4ca9b3859 mm/madvise: intro... |
54 55 |
case MADV_POPULATE_READ: case MADV_POPULATE_WRITE: |
0a27a14a6 mm: madvise avoid... |
56 57 58 59 60 61 62 63 |
return 0; default: /* be safe, default to 1. list exceptions explicitly */ return 1; } } /* |
1da177e4c Linux-2.6.12-rc2 |
64 65 66 |
* We can potentially split a vm area into separate * areas, each area with its own behavior. */ |
ec9bed9d3 mm/madvise.c: fix... |
67 |
static long madvise_behavior(struct vm_area_struct *vma, |
05b743847 [PATCH] madvise: ... |
68 69 |
struct vm_area_struct **prev, unsigned long start, unsigned long end, int behavior) |
1da177e4c Linux-2.6.12-rc2 |
70 |
{ |
ec9bed9d3 mm/madvise.c: fix... |
71 |
struct mm_struct *mm = vma->vm_mm; |
1da177e4c Linux-2.6.12-rc2 |
72 |
int error = 0; |
05b743847 [PATCH] madvise: ... |
73 |
pgoff_t pgoff; |
3866ea90d ksm: first tidy u... |
74 |
unsigned long new_flags = vma->vm_flags; |
e798c6e87 [PATCH] madvise: ... |
75 76 |
switch (behavior) { |
f82256616 [PATCH] madvise M... |
77 78 79 |
case MADV_NORMAL: new_flags = new_flags & ~VM_RAND_READ & ~VM_SEQ_READ; break; |
e798c6e87 [PATCH] madvise: ... |
80 |
case MADV_SEQUENTIAL: |
f82256616 [PATCH] madvise M... |
81 |
new_flags = (new_flags & ~VM_RAND_READ) | VM_SEQ_READ; |
e798c6e87 [PATCH] madvise: ... |
82 83 |
break; case MADV_RANDOM: |
f82256616 [PATCH] madvise M... |
84 |
new_flags = (new_flags & ~VM_SEQ_READ) | VM_RAND_READ; |
e798c6e87 [PATCH] madvise: ... |
85 |
break; |
f82256616 [PATCH] madvise M... |
86 87 88 89 |
case MADV_DONTFORK: new_flags |= VM_DONTCOPY; break; case MADV_DOFORK: |
3866ea90d ksm: first tidy u... |
90 91 92 93 |
if (vma->vm_flags & VM_IO) { error = -EINVAL; goto out; } |
f82256616 [PATCH] madvise M... |
94 |
new_flags &= ~VM_DONTCOPY; |
e798c6e87 [PATCH] madvise: ... |
95 |
break; |
d2cd9ede6 mm,fork: introduc... |
96 97 98 99 100 101 102 103 104 105 106 |
case MADV_WIPEONFORK: /* MADV_WIPEONFORK is only supported on anonymous memory. */ if (vma->vm_file || vma->vm_flags & VM_SHARED) { error = -EINVAL; goto out; } new_flags |= VM_WIPEONFORK; break; case MADV_KEEPONFORK: new_flags &= ~VM_WIPEONFORK; break; |
accb61fe7 coredump: add VM_... |
107 |
case MADV_DONTDUMP: |
0103bd16f mm: prepare VM_DO... |
108 |
new_flags |= VM_DONTDUMP; |
accb61fe7 coredump: add VM_... |
109 110 |
break; case MADV_DODUMP: |
d41aa5252 mm: madvise(MADV_... |
111 |
if (!is_vm_hugetlb_page(vma) && new_flags & VM_SPECIAL) { |
0103bd16f mm: prepare VM_DO... |
112 113 114 115 |
error = -EINVAL; goto out; } new_flags &= ~VM_DONTDUMP; |
accb61fe7 coredump: add VM_... |
116 |
break; |
f8af4da3b ksm: the mm inter... |
117 118 119 |
case MADV_MERGEABLE: case MADV_UNMERGEABLE: error = ksm_madvise(vma, start, end, behavior, &new_flags); |
f3bc0dba3 mm/madvise: reduc... |
120 121 |
if (error) goto out_convert_errno; |
f8af4da3b ksm: the mm inter... |
122 |
break; |
0af4e98b6 thp: madvise(MADV... |
123 |
case MADV_HUGEPAGE: |
a664b2d85 thp: madvise(MADV... |
124 |
case MADV_NOHUGEPAGE: |
60ab3244e thp: khugepaged: ... |
125 |
error = hugepage_madvise(vma, &new_flags, behavior); |
f3bc0dba3 mm/madvise: reduc... |
126 127 |
if (error) goto out_convert_errno; |
0af4e98b6 thp: madvise(MADV... |
128 |
break; |
e798c6e87 [PATCH] madvise: ... |
129 |
} |
05b743847 [PATCH] madvise: ... |
130 131 |
if (new_flags == vma->vm_flags) { *prev = vma; |
836d5ffd3 [PATCH] mm: fix m... |
132 |
goto out; |
05b743847 [PATCH] madvise: ... |
133 134 135 136 |
} pgoff = vma->vm_pgoff + ((start - vma->vm_start) >> PAGE_SHIFT); *prev = vma_merge(mm, *prev, start, end, new_flags, vma->anon_vma, |
19a809afe userfaultfd: teac... |
137 138 |
vma->vm_file, pgoff, vma_policy(vma), vma->vm_userfaultfd_ctx); |
05b743847 [PATCH] madvise: ... |
139 140 141 142 143 144 |
if (*prev) { vma = *prev; goto success; } *prev = vma; |
1da177e4c Linux-2.6.12-rc2 |
145 146 |
if (start != vma->vm_start) { |
def5efe03 mm, madvise: fail... |
147 148 |
if (unlikely(mm->map_count >= sysctl_max_map_count)) { error = -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
149 |
goto out; |
def5efe03 mm, madvise: fail... |
150 151 |
} error = __split_vma(mm, vma, start, 1); |
f3bc0dba3 mm/madvise: reduc... |
152 153 |
if (error) goto out_convert_errno; |
1da177e4c Linux-2.6.12-rc2 |
154 155 156 |
} if (end != vma->vm_end) { |
def5efe03 mm, madvise: fail... |
157 158 |
if (unlikely(mm->map_count >= sysctl_max_map_count)) { error = -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
159 |
goto out; |
def5efe03 mm, madvise: fail... |
160 161 |
} error = __split_vma(mm, vma, end, 0); |
f3bc0dba3 mm/madvise: reduc... |
162 163 |
if (error) goto out_convert_errno; |
1da177e4c Linux-2.6.12-rc2 |
164 |
} |
836d5ffd3 [PATCH] mm: fix m... |
165 |
success: |
1da177e4c Linux-2.6.12-rc2 |
166 |
/* |
c1e8d7c6a mmap locking API:... |
167 |
* vm_flags is protected by the mmap_lock held in write mode. |
1da177e4c Linux-2.6.12-rc2 |
168 |
*/ |
e798c6e87 [PATCH] madvise: ... |
169 |
vma->vm_flags = new_flags; |
f3bc0dba3 mm/madvise: reduc... |
170 171 172 173 174 175 176 177 |
out_convert_errno: /* * madvise() returns EAGAIN if kernel resources, such as * slab, are temporarily unavailable. */ if (error == -ENOMEM) error = -EAGAIN; |
1da177e4c Linux-2.6.12-rc2 |
178 |
out: |
1da177e4c Linux-2.6.12-rc2 |
179 180 |
return error; } |
1998cc048 mm: make madvise(... |
181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 |
#ifdef CONFIG_SWAP static int swapin_walk_pmd_entry(pmd_t *pmd, unsigned long start, unsigned long end, struct mm_walk *walk) { pte_t *orig_pte; struct vm_area_struct *vma = walk->private; unsigned long index; if (pmd_none_or_trans_huge_or_clear_bad(pmd)) return 0; for (index = start; index != end; index += PAGE_SIZE) { pte_t pte; swp_entry_t entry; struct page *page; spinlock_t *ptl; orig_pte = pte_offset_map_lock(vma->vm_mm, pmd, start, &ptl); pte = *(orig_pte + ((index - start) / PAGE_SIZE)); pte_unmap_unlock(orig_pte, ptl); |
0661a3361 mm: remove rest u... |
201 |
if (pte_present(pte) || pte_none(pte)) |
1998cc048 mm: make madvise(... |
202 203 204 205 206 207 |
continue; entry = pte_to_swp_entry(pte); if (unlikely(non_swap_entry(entry))) continue; page = read_swap_cache_async(entry, GFP_HIGHUSER_MOVABLE, |
23955622f swap: add block i... |
208 |
vma, index, false); |
1998cc048 mm: make madvise(... |
209 |
if (page) |
09cbfeaf1 mm, fs: get rid o... |
210 |
put_page(page); |
1998cc048 mm: make madvise(... |
211 212 213 214 |
} return 0; } |
7b86ac337 pagewalk: separat... |
215 216 217 |
static const struct mm_walk_ops swapin_walk_ops = { .pmd_entry = swapin_walk_pmd_entry, }; |
1998cc048 mm: make madvise(... |
218 219 220 221 222 |
static void force_shm_swapin_readahead(struct vm_area_struct *vma, unsigned long start, unsigned long end, struct address_space *mapping) { |
e6e88712e mm: optimise madv... |
223 |
XA_STATE(xas, &mapping->i_pages, linear_page_index(vma, start)); |
66383800d mm: fix madvise W... |
224 |
pgoff_t end_index = linear_page_index(vma, end + PAGE_SIZE - 1); |
1998cc048 mm: make madvise(... |
225 |
struct page *page; |
1998cc048 mm: make madvise(... |
226 |
|
e6e88712e mm: optimise madv... |
227 228 229 |
rcu_read_lock(); xas_for_each(&xas, page, end_index) { swp_entry_t swap; |
1998cc048 mm: make madvise(... |
230 |
|
e6e88712e mm: optimise madv... |
231 |
if (!xa_is_value(page)) |
1998cc048 mm: make madvise(... |
232 |
continue; |
e6e88712e mm: optimise madv... |
233 234 |
xas_pause(&xas); rcu_read_unlock(); |
1998cc048 mm: make madvise(... |
235 236 |
swap = radix_to_swp_entry(page); page = read_swap_cache_async(swap, GFP_HIGHUSER_MOVABLE, |
23955622f swap: add block i... |
237 |
NULL, 0, false); |
1998cc048 mm: make madvise(... |
238 |
if (page) |
09cbfeaf1 mm, fs: get rid o... |
239 |
put_page(page); |
e6e88712e mm: optimise madv... |
240 241 |
rcu_read_lock(); |
1998cc048 mm: make madvise(... |
242 |
} |
e6e88712e mm: optimise madv... |
243 |
rcu_read_unlock(); |
1998cc048 mm: make madvise(... |
244 245 246 247 |
lru_add_drain(); /* Push any new pages onto the LRU now */ } #endif /* CONFIG_SWAP */ |
1da177e4c Linux-2.6.12-rc2 |
248 249 250 |
/* * Schedule all required I/O operations. Do not wait for completion. */ |
ec9bed9d3 mm/madvise.c: fix... |
251 252 |
static long madvise_willneed(struct vm_area_struct *vma, struct vm_area_struct **prev, |
1da177e4c Linux-2.6.12-rc2 |
253 254 |
unsigned long start, unsigned long end) { |
0726b01e7 mm/madvise: pass ... |
255 |
struct mm_struct *mm = vma->vm_mm; |
1da177e4c Linux-2.6.12-rc2 |
256 |
struct file *file = vma->vm_file; |
692fe6243 mm: Handle MADV_W... |
257 |
loff_t offset; |
1da177e4c Linux-2.6.12-rc2 |
258 |
|
6ea8d958a mm/madvise.c: fix... |
259 |
*prev = vma; |
1998cc048 mm: make madvise(... |
260 |
#ifdef CONFIG_SWAP |
97b713ba3 fs: kill BDI_CAP_... |
261 |
if (!file) { |
7b86ac337 pagewalk: separat... |
262 263 |
walk_page_range(vma->vm_mm, start, end, &swapin_walk_ops, vma); lru_add_drain(); /* Push any new pages onto the LRU now */ |
1998cc048 mm: make madvise(... |
264 265 |
return 0; } |
1998cc048 mm: make madvise(... |
266 |
|
97b713ba3 fs: kill BDI_CAP_... |
267 |
if (shmem_mapping(file->f_mapping)) { |
97b713ba3 fs: kill BDI_CAP_... |
268 269 270 271 272 |
force_shm_swapin_readahead(vma, start, end, file->f_mapping); return 0; } #else |
1bef40032 [PATCH] madvise: ... |
273 274 |
if (!file) return -EBADF; |
97b713ba3 fs: kill BDI_CAP_... |
275 |
#endif |
1bef40032 [PATCH] madvise: ... |
276 |
|
e748dcd09 vfs: remove get_x... |
277 |
if (IS_DAX(file_inode(file))) { |
fe77ba6f4 [PATCH] xip: madv... |
278 279 280 |
/* no bad return value, but ignore advice */ return 0; } |
692fe6243 mm: Handle MADV_W... |
281 282 283 284 |
/* * Filesystem's fadvise may need to take various locks. We need to * explicitly grab a reference because the vma (and hence the * vma's reference to the file) can go away as soon as we drop |
c1e8d7c6a mmap locking API:... |
285 |
* mmap_lock. |
692fe6243 mm: Handle MADV_W... |
286 |
*/ |
c1e8d7c6a mmap locking API:... |
287 |
*prev = NULL; /* tell sys_madvise we drop mmap_lock */ |
692fe6243 mm: Handle MADV_W... |
288 |
get_file(file); |
692fe6243 mm: Handle MADV_W... |
289 290 |
offset = (loff_t)(start - vma->vm_start) + ((loff_t)vma->vm_pgoff << PAGE_SHIFT); |
0726b01e7 mm/madvise: pass ... |
291 |
mmap_read_unlock(mm); |
692fe6243 mm: Handle MADV_W... |
292 293 |
vfs_fadvise(file, offset, end - start, POSIX_FADV_WILLNEED); fput(file); |
0726b01e7 mm/madvise: pass ... |
294 |
mmap_read_lock(mm); |
1da177e4c Linux-2.6.12-rc2 |
295 296 |
return 0; } |
d616d5126 mm: factor out co... |
297 298 299 |
static int madvise_cold_or_pageout_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end, struct mm_walk *walk) |
9c276cc65 mm: introduce MAD... |
300 |
{ |
d616d5126 mm: factor out co... |
301 302 303 |
struct madvise_walk_private *private = walk->private; struct mmu_gather *tlb = private->tlb; bool pageout = private->pageout; |
9c276cc65 mm: introduce MAD... |
304 305 306 307 |
struct mm_struct *mm = tlb->mm; struct vm_area_struct *vma = walk->vma; pte_t *orig_pte, *pte, ptent; spinlock_t *ptl; |
d616d5126 mm: factor out co... |
308 309 310 311 312 |
struct page *page = NULL; LIST_HEAD(page_list); if (fatal_signal_pending(current)) return -EINTR; |
9c276cc65 mm: introduce MAD... |
313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 |
#ifdef CONFIG_TRANSPARENT_HUGEPAGE if (pmd_trans_huge(*pmd)) { pmd_t orig_pmd; unsigned long next = pmd_addr_end(addr, end); tlb_change_page_size(tlb, HPAGE_PMD_SIZE); ptl = pmd_trans_huge_lock(pmd, vma); if (!ptl) return 0; orig_pmd = *pmd; if (is_huge_zero_pmd(orig_pmd)) goto huge_unlock; if (unlikely(!pmd_present(orig_pmd))) { VM_BUG_ON(thp_migration_supported() && !is_pmd_migration_entry(orig_pmd)); goto huge_unlock; } page = pmd_page(orig_pmd); |
12e967fd8 mm: do not allow ... |
335 336 337 338 |
/* Do not interfere with other mappings of this page */ if (page_mapcount(page) != 1) goto huge_unlock; |
9c276cc65 mm: introduce MAD... |
339 340 |
if (next - addr != HPAGE_PMD_SIZE) { int err; |
9c276cc65 mm: introduce MAD... |
341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 |
get_page(page); spin_unlock(ptl); lock_page(page); err = split_huge_page(page); unlock_page(page); put_page(page); if (!err) goto regular_page; return 0; } if (pmd_young(orig_pmd)) { pmdp_invalidate(vma, addr, pmd); orig_pmd = pmd_mkold(orig_pmd); set_pmd_at(mm, addr, pmd, orig_pmd); tlb_remove_pmd_tlb_entry(tlb, pmd, addr); } |
d616d5126 mm: factor out co... |
359 |
ClearPageReferenced(page); |
9c276cc65 mm: introduce MAD... |
360 |
test_and_clear_page_young(page); |
d616d5126 mm: factor out co... |
361 |
if (pageout) { |
820729629 mm: fix trying to... |
362 363 364 365 366 367 |
if (!isolate_lru_page(page)) { if (PageUnevictable(page)) putback_lru_page(page); else list_add(&page->lru, &page_list); } |
d616d5126 mm: factor out co... |
368 369 |
} else deactivate_page(page); |
9c276cc65 mm: introduce MAD... |
370 371 |
huge_unlock: spin_unlock(ptl); |
d616d5126 mm: factor out co... |
372 373 |
if (pageout) reclaim_pages(&page_list); |
9c276cc65 mm: introduce MAD... |
374 375 |
return 0; } |
ce2684254 mm: validate pmd ... |
376 |
regular_page: |
9c276cc65 mm: introduce MAD... |
377 378 |
if (pmd_trans_unstable(pmd)) return 0; |
9c276cc65 mm: introduce MAD... |
379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 |
#endif tlb_change_page_size(tlb, PAGE_SIZE); orig_pte = pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl); flush_tlb_batched_pending(mm); arch_enter_lazy_mmu_mode(); for (; addr < end; pte++, addr += PAGE_SIZE) { ptent = *pte; if (pte_none(ptent)) continue; if (!pte_present(ptent)) continue; page = vm_normal_page(vma, addr, ptent); if (!page) continue; /* * Creating a THP page is expensive so split it only if we * are sure it's worth. Split it if we are only owner. */ if (PageTransCompound(page)) { if (page_mapcount(page) != 1) break; get_page(page); if (!trylock_page(page)) { put_page(page); break; } pte_unmap_unlock(orig_pte, ptl); if (split_huge_page(page)) { unlock_page(page); put_page(page); pte_offset_map_lock(mm, pmd, addr, &ptl); break; } unlock_page(page); put_page(page); pte = pte_offset_map_lock(mm, pmd, addr, &ptl); pte--; addr -= PAGE_SIZE; continue; } |
12e967fd8 mm: do not allow ... |
423 424 425 |
/* Do not interfere with other mappings of this page */ if (page_mapcount(page) != 1) continue; |
9c276cc65 mm: introduce MAD... |
426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 |
VM_BUG_ON_PAGE(PageTransCompound(page), page); if (pte_young(ptent)) { ptent = ptep_get_and_clear_full(mm, addr, pte, tlb->fullmm); ptent = pte_mkold(ptent); set_pte_at(mm, addr, pte, ptent); tlb_remove_tlb_entry(tlb, pte, addr); } /* * We are deactivating a page for accelerating reclaiming. * VM couldn't reclaim the page unless we clear PG_young. * As a side effect, it makes confuse idle-page tracking * because they will miss recent referenced history. */ |
d616d5126 mm: factor out co... |
442 |
ClearPageReferenced(page); |
9c276cc65 mm: introduce MAD... |
443 |
test_and_clear_page_young(page); |
d616d5126 mm: factor out co... |
444 |
if (pageout) { |
820729629 mm: fix trying to... |
445 446 447 448 449 450 |
if (!isolate_lru_page(page)) { if (PageUnevictable(page)) putback_lru_page(page); else list_add(&page->lru, &page_list); } |
d616d5126 mm: factor out co... |
451 452 |
} else deactivate_page(page); |
9c276cc65 mm: introduce MAD... |
453 454 455 456 |
} arch_leave_lazy_mmu_mode(); pte_unmap_unlock(orig_pte, ptl); |
d616d5126 mm: factor out co... |
457 458 |
if (pageout) reclaim_pages(&page_list); |
9c276cc65 mm: introduce MAD... |
459 460 461 462 463 464 |
cond_resched(); return 0; } static const struct mm_walk_ops cold_walk_ops = { |
d616d5126 mm: factor out co... |
465 |
.pmd_entry = madvise_cold_or_pageout_pte_range, |
9c276cc65 mm: introduce MAD... |
466 467 468 469 470 471 |
}; static void madvise_cold_page_range(struct mmu_gather *tlb, struct vm_area_struct *vma, unsigned long addr, unsigned long end) { |
d616d5126 mm: factor out co... |
472 473 474 475 |
struct madvise_walk_private walk_private = { .pageout = false, .tlb = tlb, }; |
9c276cc65 mm: introduce MAD... |
476 |
tlb_start_vma(tlb, vma); |
d616d5126 mm: factor out co... |
477 |
walk_page_range(vma->vm_mm, addr, end, &cold_walk_ops, &walk_private); |
9c276cc65 mm: introduce MAD... |
478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 |
tlb_end_vma(tlb, vma); } static long madvise_cold(struct vm_area_struct *vma, struct vm_area_struct **prev, unsigned long start_addr, unsigned long end_addr) { struct mm_struct *mm = vma->vm_mm; struct mmu_gather tlb; *prev = vma; if (!can_madv_lru_vma(vma)) return -EINVAL; lru_add_drain(); |
a72afd873 tlb: mmu_gather: ... |
493 |
tlb_gather_mmu(&tlb, mm); |
9c276cc65 mm: introduce MAD... |
494 |
madvise_cold_page_range(&tlb, vma, start_addr, end_addr); |
ae8eba8b5 tlb: mmu_gather: ... |
495 |
tlb_finish_mmu(&tlb); |
9c276cc65 mm: introduce MAD... |
496 497 498 |
return 0; } |
1a4e58cce mm: introduce MAD... |
499 500 501 502 |
static void madvise_pageout_page_range(struct mmu_gather *tlb, struct vm_area_struct *vma, unsigned long addr, unsigned long end) { |
d616d5126 mm: factor out co... |
503 504 505 506 |
struct madvise_walk_private walk_private = { .pageout = true, .tlb = tlb, }; |
1a4e58cce mm: introduce MAD... |
507 |
tlb_start_vma(tlb, vma); |
d616d5126 mm: factor out co... |
508 |
walk_page_range(vma->vm_mm, addr, end, &cold_walk_ops, &walk_private); |
1a4e58cce mm: introduce MAD... |
509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 |
tlb_end_vma(tlb, vma); } static inline bool can_do_pageout(struct vm_area_struct *vma) { if (vma_is_anonymous(vma)) return true; if (!vma->vm_file) return false; /* * paging out pagecache only for non-anonymous mappings that correspond * to the files the calling process could (if tried) open for writing; * otherwise we'd be including shared non-exclusive mappings, which * opens a side channel. */ |
21cb47be6 inode: make init ... |
524 525 |
return inode_owner_or_capable(&init_user_ns, file_inode(vma->vm_file)) || |
02f92b386 fs: add file and ... |
526 |
file_permission(vma->vm_file, MAY_WRITE) == 0; |
1a4e58cce mm: introduce MAD... |
527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 |
} static long madvise_pageout(struct vm_area_struct *vma, struct vm_area_struct **prev, unsigned long start_addr, unsigned long end_addr) { struct mm_struct *mm = vma->vm_mm; struct mmu_gather tlb; *prev = vma; if (!can_madv_lru_vma(vma)) return -EINVAL; if (!can_do_pageout(vma)) return 0; lru_add_drain(); |
a72afd873 tlb: mmu_gather: ... |
544 |
tlb_gather_mmu(&tlb, mm); |
1a4e58cce mm: introduce MAD... |
545 |
madvise_pageout_page_range(&tlb, vma, start_addr, end_addr); |
ae8eba8b5 tlb: mmu_gather: ... |
546 |
tlb_finish_mmu(&tlb); |
1a4e58cce mm: introduce MAD... |
547 548 549 |
return 0; } |
854e9ed09 mm: support madvi... |
550 551 552 553 554 555 556 557 558 559 |
static int madvise_free_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end, struct mm_walk *walk) { struct mmu_gather *tlb = walk->private; struct mm_struct *mm = tlb->mm; struct vm_area_struct *vma = walk->vma; spinlock_t *ptl; pte_t *orig_pte, *pte, ptent; struct page *page; |
64b42bc1c mm/madvise.c: fre... |
560 |
int nr_swap = 0; |
b8d3c4c30 mm/huge_memory.c:... |
561 562 563 564 565 566 |
unsigned long next; next = pmd_addr_end(addr, end); if (pmd_trans_huge(*pmd)) if (madvise_free_huge_pmd(tlb, vma, pmd, addr, next)) goto next; |
854e9ed09 mm: support madvi... |
567 |
|
854e9ed09 mm: support madvi... |
568 569 |
if (pmd_trans_unstable(pmd)) return 0; |
ed6a79352 asm-generic/tlb, ... |
570 |
tlb_change_page_size(tlb, PAGE_SIZE); |
854e9ed09 mm: support madvi... |
571 |
orig_pte = pte = pte_offset_map_lock(mm, pmd, addr, &ptl); |
3ea277194 mm, mprotect: flu... |
572 |
flush_tlb_batched_pending(mm); |
854e9ed09 mm: support madvi... |
573 574 575 |
arch_enter_lazy_mmu_mode(); for (; addr != end; pte++, addr += PAGE_SIZE) { ptent = *pte; |
64b42bc1c mm/madvise.c: fre... |
576 |
if (pte_none(ptent)) |
854e9ed09 mm: support madvi... |
577 |
continue; |
64b42bc1c mm/madvise.c: fre... |
578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 |
/* * If the pte has swp_entry, just clear page table to * prevent swap-in which is more expensive rather than * (page allocation + zeroing). */ if (!pte_present(ptent)) { swp_entry_t entry; entry = pte_to_swp_entry(ptent); if (non_swap_entry(entry)) continue; nr_swap--; free_swap_and_cache(entry); pte_clear_not_present_full(mm, addr, pte, tlb->fullmm); continue; } |
854e9ed09 mm: support madvi... |
594 |
|
25b2995a3 mm: remove MEMORY... |
595 |
page = vm_normal_page(vma, addr, ptent); |
854e9ed09 mm: support madvi... |
596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 |
if (!page) continue; /* * If pmd isn't transhuge but the page is THP and * is owned by only this process, split it and * deactivate all pages. */ if (PageTransCompound(page)) { if (page_mapcount(page) != 1) goto out; get_page(page); if (!trylock_page(page)) { put_page(page); goto out; } pte_unmap_unlock(orig_pte, ptl); if (split_huge_page(page)) { unlock_page(page); put_page(page); pte_offset_map_lock(mm, pmd, addr, &ptl); goto out; } |
854e9ed09 mm: support madvi... |
619 |
unlock_page(page); |
263630e8d mm/madvise.c: fix... |
620 |
put_page(page); |
854e9ed09 mm: support madvi... |
621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 |
pte = pte_offset_map_lock(mm, pmd, addr, &ptl); pte--; addr -= PAGE_SIZE; continue; } VM_BUG_ON_PAGE(PageTransCompound(page), page); if (PageSwapCache(page) || PageDirty(page)) { if (!trylock_page(page)) continue; /* * If page is shared with others, we couldn't clear * PG_dirty of the page. */ if (page_mapcount(page) != 1) { unlock_page(page); continue; } if (PageSwapCache(page) && !try_to_free_swap(page)) { unlock_page(page); continue; } ClearPageDirty(page); unlock_page(page); } if (pte_young(ptent) || pte_dirty(ptent)) { /* * Some of architecture(ex, PPC) don't update TLB * with set_pte_at and tlb_remove_tlb_entry so for * the portability, remap the pte with old|clean * after pte clearing. */ ptent = ptep_get_and_clear_full(mm, addr, pte, tlb->fullmm); ptent = pte_mkold(ptent); ptent = pte_mkclean(ptent); set_pte_at(mm, addr, pte, ptent); tlb_remove_tlb_entry(tlb, pte, addr); } |
802a3a92a mm: reclaim MADV_... |
665 |
mark_page_lazyfree(page); |
854e9ed09 mm: support madvi... |
666 667 |
} out: |
64b42bc1c mm/madvise.c: fre... |
668 669 670 671 672 673 |
if (nr_swap) { if (current->mm == mm) sync_mm_rss(mm); add_mm_counter(mm, MM_SWAPENTS, nr_swap); } |
854e9ed09 mm: support madvi... |
674 675 676 |
arch_leave_lazy_mmu_mode(); pte_unmap_unlock(orig_pte, ptl); cond_resched(); |
b8d3c4c30 mm/huge_memory.c:... |
677 |
next: |
854e9ed09 mm: support madvi... |
678 679 |
return 0; } |
7b86ac337 pagewalk: separat... |
680 681 682 |
static const struct mm_walk_ops madvise_free_walk_ops = { .pmd_entry = madvise_free_pte_range, }; |
854e9ed09 mm: support madvi... |
683 684 685 686 |
static int madvise_free_single_vma(struct vm_area_struct *vma, unsigned long start_addr, unsigned long end_addr) { |
854e9ed09 mm: support madvi... |
687 |
struct mm_struct *mm = vma->vm_mm; |
ac46d4f3c mm/mmu_notifier: ... |
688 |
struct mmu_notifier_range range; |
854e9ed09 mm: support madvi... |
689 |
struct mmu_gather tlb; |
854e9ed09 mm: support madvi... |
690 691 692 |
/* MADV_FREE works for only anon vma at the moment */ if (!vma_is_anonymous(vma)) return -EINVAL; |
ac46d4f3c mm/mmu_notifier: ... |
693 694 |
range.start = max(vma->vm_start, start_addr); if (range.start >= vma->vm_end) |
854e9ed09 mm: support madvi... |
695 |
return -EINVAL; |
ac46d4f3c mm/mmu_notifier: ... |
696 697 |
range.end = min(vma->vm_end, end_addr); if (range.end <= vma->vm_start) |
854e9ed09 mm: support madvi... |
698 |
return -EINVAL; |
7269f9999 mm/mmu_notifier: ... |
699 |
mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, vma, mm, |
6f4f13e8d mm/mmu_notifier: ... |
700 |
range.start, range.end); |
854e9ed09 mm: support madvi... |
701 702 |
lru_add_drain(); |
a72afd873 tlb: mmu_gather: ... |
703 |
tlb_gather_mmu(&tlb, mm); |
854e9ed09 mm: support madvi... |
704 |
update_hiwater_rss(mm); |
ac46d4f3c mm/mmu_notifier: ... |
705 |
mmu_notifier_invalidate_range_start(&range); |
7b86ac337 pagewalk: separat... |
706 707 708 709 |
tlb_start_vma(&tlb, vma); walk_page_range(vma->vm_mm, range.start, range.end, &madvise_free_walk_ops, &tlb); tlb_end_vma(&tlb, vma); |
ac46d4f3c mm/mmu_notifier: ... |
710 |
mmu_notifier_invalidate_range_end(&range); |
ae8eba8b5 tlb: mmu_gather: ... |
711 |
tlb_finish_mmu(&tlb); |
854e9ed09 mm: support madvi... |
712 713 714 |
return 0; } |
1da177e4c Linux-2.6.12-rc2 |
715 716 717 718 |
/* * Application no longer needs these pages. If the pages are dirty, * it's OK to just throw them away. The app will be more careful about * data it wants to keep. Be sure to free swap resources too. The |
7e6cbea39 madvise: update f... |
719 |
* zap_page_range call sets things up for shrink_active_list to actually free |
1da177e4c Linux-2.6.12-rc2 |
720 721 |
* these pages later if no one else has touched them in the meantime, * although we could add these pages to a global reuse list for |
7e6cbea39 madvise: update f... |
722 |
* shrink_active_list to pick up before reclaiming other pages. |
1da177e4c Linux-2.6.12-rc2 |
723 724 725 726 727 728 729 730 731 732 733 |
* * NB: This interface discards data rather than pushes it out to swap, * as some implementations do. This has performance implications for * applications like large transactional databases which want to discard * pages in anonymous maps after committing to backing store the data * that was kept in them. There is no reason to write this data out to * the swap area if the application is discarding it. * * An interface that causes the system to free clean pages and flush * dirty pages is already available as msync(MS_INVALIDATE). */ |
230ca982b userfaultfd: non-... |
734 735 736 737 738 739 740 741 742 743 744 |
static long madvise_dontneed_single_vma(struct vm_area_struct *vma, unsigned long start, unsigned long end) { zap_page_range(vma, start, end - start); return 0; } static long madvise_dontneed_free(struct vm_area_struct *vma, struct vm_area_struct **prev, unsigned long start, unsigned long end, int behavior) |
1da177e4c Linux-2.6.12-rc2 |
745 |
{ |
0726b01e7 mm/madvise: pass ... |
746 |
struct mm_struct *mm = vma->vm_mm; |
05b743847 [PATCH] madvise: ... |
747 |
*prev = vma; |
9c276cc65 mm: introduce MAD... |
748 |
if (!can_madv_lru_vma(vma)) |
1da177e4c Linux-2.6.12-rc2 |
749 |
return -EINVAL; |
70ccb92fd userfaultfd: non-... |
750 |
if (!userfaultfd_remove(vma, start, end)) { |
c1e8d7c6a mmap locking API:... |
751 |
*prev = NULL; /* mmap_lock has been dropped, prev is stale */ |
70ccb92fd userfaultfd: non-... |
752 |
|
0726b01e7 mm/madvise: pass ... |
753 754 |
mmap_read_lock(mm); vma = find_vma(mm, start); |
70ccb92fd userfaultfd: non-... |
755 756 757 758 759 760 761 762 763 |
if (!vma) return -ENOMEM; if (start < vma->vm_start) { /* * This "vma" under revalidation is the one * with the lowest vma->vm_start where start * is also < vma->vm_end. If start < * vma->vm_start it means an hole materialized * in the user address space within the |
230ca982b userfaultfd: non-... |
764 765 |
* virtual range passed to MADV_DONTNEED * or MADV_FREE. |
70ccb92fd userfaultfd: non-... |
766 767 768 |
*/ return -ENOMEM; } |
9c276cc65 mm: introduce MAD... |
769 |
if (!can_madv_lru_vma(vma)) |
70ccb92fd userfaultfd: non-... |
770 771 772 773 |
return -EINVAL; if (end > vma->vm_end) { /* * Don't fail if end > vma->vm_end. If the old |
f0953a1bb mm: fix typos in ... |
774 |
* vma was split while the mmap_lock was |
70ccb92fd userfaultfd: non-... |
775 |
* released the effect of the concurrent |
230ca982b userfaultfd: non-... |
776 |
* operation may not cause madvise() to |
70ccb92fd userfaultfd: non-... |
777 778 779 780 781 782 783 784 785 786 787 |
* have an undefined result. There may be an * adjacent next vma that we'll walk * next. userfaultfd_remove() will generate an * UFFD_EVENT_REMOVE repetition on the * end-vma->vm_end range, but the manager can * handle a repetition fine. */ end = vma->vm_end; } VM_WARN_ON(start >= end); } |
230ca982b userfaultfd: non-... |
788 789 790 791 792 793 794 |
if (behavior == MADV_DONTNEED) return madvise_dontneed_single_vma(vma, start, end); else if (behavior == MADV_FREE) return madvise_free_single_vma(vma, start, end); else return -EINVAL; |
1da177e4c Linux-2.6.12-rc2 |
795 |
} |
4ca9b3859 mm/madvise: intro... |
796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 |
static long madvise_populate(struct vm_area_struct *vma, struct vm_area_struct **prev, unsigned long start, unsigned long end, int behavior) { const bool write = behavior == MADV_POPULATE_WRITE; struct mm_struct *mm = vma->vm_mm; unsigned long tmp_end; int locked = 1; long pages; *prev = vma; while (start < end) { /* * We might have temporarily dropped the lock. For example, * our VMA might have been split. */ if (!vma || start >= vma->vm_end) { vma = find_vma(mm, start); if (!vma || start < vma->vm_start) return -ENOMEM; } tmp_end = min_t(unsigned long, end, vma->vm_end); /* Populate (prefault) page tables readable/writable. */ pages = faultin_vma_page_range(vma, start, tmp_end, write, &locked); if (!locked) { mmap_read_lock(mm); locked = 1; *prev = NULL; vma = NULL; } if (pages < 0) { switch (pages) { case -EINTR: return -EINTR; |
eb2faa513 mm/madvise: repor... |
834 |
case -EINVAL: /* Incompatible mappings / permissions. */ |
4ca9b3859 mm/madvise: intro... |
835 836 837 |
return -EINVAL; case -EHWPOISON: return -EHWPOISON; |
eb2faa513 mm/madvise: repor... |
838 839 |
case -EFAULT: /* VM_FAULT_SIGBUS or VM_FAULT_SIGSEGV */ return -EFAULT; |
4ca9b3859 mm/madvise: intro... |
840 841 842 843 844 845 846 847 848 849 850 851 852 |
default: pr_warn_once("%s: unhandled return value: %ld ", __func__, pages); fallthrough; case -ENOMEM: return -ENOMEM; } } start += pages * PAGE_SIZE; } return 0; } |
f6b3ec238 [PATCH] madvise(M... |
853 854 855 |
/* * Application wants to free up the pages and associated backing store. * This is effectively punching a hole into the middle of a file. |
f6b3ec238 [PATCH] madvise(M... |
856 857 |
*/ static long madvise_remove(struct vm_area_struct *vma, |
00e9fa2d6 [PATCH] mm: fix m... |
858 |
struct vm_area_struct **prev, |
f6b3ec238 [PATCH] madvise(M... |
859 860 |
unsigned long start, unsigned long end) { |
3f31d0757 mm/fs: route MADV... |
861 |
loff_t offset; |
90ed52ebe [PATCH] holepunch... |
862 |
int error; |
9ab4233dd mm: Hold a file r... |
863 |
struct file *f; |
0726b01e7 mm/madvise: pass ... |
864 |
struct mm_struct *mm = vma->vm_mm; |
f6b3ec238 [PATCH] madvise(M... |
865 |
|
c1e8d7c6a mmap locking API:... |
866 |
*prev = NULL; /* tell sys_madvise we drop mmap_lock */ |
00e9fa2d6 [PATCH] mm: fix m... |
867 |
|
72079ba0d mm: madvise allow... |
868 |
if (vma->vm_flags & VM_LOCKED) |
f6b3ec238 [PATCH] madvise(M... |
869 |
return -EINVAL; |
9ab4233dd mm: Hold a file r... |
870 871 872 |
f = vma->vm_file; if (!f || !f->f_mapping || !f->f_mapping->host) { |
f6b3ec238 [PATCH] madvise(M... |
873 874 |
return -EINVAL; } |
69cf0fac6 [PATCH] Fix MADV_... |
875 876 |
if ((vma->vm_flags & (VM_SHARED|VM_WRITE)) != (VM_SHARED|VM_WRITE)) return -EACCES; |
f6b3ec238 [PATCH] madvise(M... |
877 878 |
offset = (loff_t)(start - vma->vm_start) + ((loff_t)vma->vm_pgoff << PAGE_SHIFT); |
90ed52ebe [PATCH] holepunch... |
879 |
|
9ab4233dd mm: Hold a file r... |
880 |
/* |
9608703e4 mm: Fix comments ... |
881 |
* Filesystem's fallocate may need to take i_rwsem. We need to |
9ab4233dd mm: Hold a file r... |
882 883 |
* explicitly grab a reference because the vma (and hence the * vma's reference to the file) can go away as soon as we drop |
c1e8d7c6a mmap locking API:... |
884 |
* mmap_lock. |
9ab4233dd mm: Hold a file r... |
885 886 |
*/ get_file(f); |
70ccb92fd userfaultfd: non-... |
887 |
if (userfaultfd_remove(vma, start, end)) { |
c1e8d7c6a mmap locking API:... |
888 |
/* mmap_lock was not released by userfaultfd_remove() */ |
0726b01e7 mm/madvise: pass ... |
889 |
mmap_read_unlock(mm); |
70ccb92fd userfaultfd: non-... |
890 |
} |
72c72bdf7 VFS: Rename do_fa... |
891 |
error = vfs_fallocate(f, |
3f31d0757 mm/fs: route MADV... |
892 893 |
FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE, offset, end - start); |
9ab4233dd mm: Hold a file r... |
894 |
fput(f); |
0726b01e7 mm/madvise: pass ... |
895 |
mmap_read_lock(mm); |
90ed52ebe [PATCH] holepunch... |
896 |
return error; |
f6b3ec238 [PATCH] madvise(M... |
897 |
} |
9893e49d6 HWPOISON: Add mad... |
898 899 900 901 |
#ifdef CONFIG_MEMORY_FAILURE /* * Error injection support for memory error handling. */ |
97167a768 mm/madvise.c: cle... |
902 903 |
static int madvise_inject_error(int behavior, unsigned long start, unsigned long end) |
9893e49d6 HWPOISON: Add mad... |
904 |
{ |
d3cd257ce mm/madvise.c: rep... |
905 |
unsigned long size; |
97167a768 mm/madvise.c: cle... |
906 |
|
9893e49d6 HWPOISON: Add mad... |
907 908 |
if (!capable(CAP_SYS_ADMIN)) return -EPERM; |
97167a768 mm/madvise.c: cle... |
909 |
|
19bfbe22f mm, hugetlb, soft... |
910 |
|
d3cd257ce mm/madvise.c: rep... |
911 |
for (; start < end; start += size) { |
23e7b5c2e mm, madvise_injec... |
912 |
unsigned long pfn; |
dc7560b49 mm,hwpoison: refa... |
913 |
struct page *page; |
325c4ef5c mm/madvise.c:madv... |
914 |
int ret; |
97167a768 mm/madvise.c: cle... |
915 |
ret = get_user_pages_fast(start, 1, 0, &page); |
9893e49d6 HWPOISON: Add mad... |
916 917 |
if (ret != 1) return ret; |
23e7b5c2e mm, madvise_injec... |
918 |
pfn = page_to_pfn(page); |
325c4ef5c mm/madvise.c:madv... |
919 |
|
19bfbe22f mm, hugetlb, soft... |
920 921 922 |
/* * When soft offlining hugepages, after migrating the page * we dissolve it, therefore in the second loop "page" will |
d3cd257ce mm/madvise.c: rep... |
923 |
* no longer be a compound page. |
19bfbe22f mm, hugetlb, soft... |
924 |
*/ |
d3cd257ce mm/madvise.c: rep... |
925 |
size = page_size(compound_head(page)); |
19bfbe22f mm, hugetlb, soft... |
926 |
|
97167a768 mm/madvise.c: cle... |
927 928 929 |
if (behavior == MADV_SOFT_OFFLINE) { pr_info("Soft offlining pfn %#lx at process virtual address %#lx ", |
dc7560b49 mm,hwpoison: refa... |
930 |
pfn, start); |
feec24a61 mm, soft-offline:... |
931 |
ret = soft_offline_page(pfn, MF_COUNT_INCREASED); |
dc7560b49 mm,hwpoison: refa... |
932 933 934 935 |
} else { pr_info("Injecting memory failure for pfn %#lx at process virtual address %#lx ", pfn, start); |
1e8aaedb1 mm,memory_failure... |
936 |
ret = memory_failure(pfn, MF_COUNT_INCREASED); |
7a07875fa mm/hwpoison: avoi... |
937 938 |
if (ret == -EOPNOTSUPP) ret = 0; |
afcf938ee HWPOISON: Add a m... |
939 |
} |
23e7b5c2e mm, madvise_injec... |
940 |
|
23a003bfd mm/madvise: pass ... |
941 942 |
if (ret) return ret; |
9893e49d6 HWPOISON: Add mad... |
943 |
} |
c461ad6a6 mm, madvise: ensu... |
944 |
|
325c4ef5c mm/madvise.c:madv... |
945 |
return 0; |
9893e49d6 HWPOISON: Add mad... |
946 947 |
} #endif |
165cd4023 [PATCH] madvise()... |
948 949 950 |
static long madvise_vma(struct vm_area_struct *vma, struct vm_area_struct **prev, unsigned long start, unsigned long end, int behavior) |
1da177e4c Linux-2.6.12-rc2 |
951 |
{ |
1da177e4c Linux-2.6.12-rc2 |
952 |
switch (behavior) { |
f6b3ec238 [PATCH] madvise(M... |
953 |
case MADV_REMOVE: |
3866ea90d ksm: first tidy u... |
954 |
return madvise_remove(vma, prev, start, end); |
1da177e4c Linux-2.6.12-rc2 |
955 |
case MADV_WILLNEED: |
3866ea90d ksm: first tidy u... |
956 |
return madvise_willneed(vma, prev, start, end); |
9c276cc65 mm: introduce MAD... |
957 958 |
case MADV_COLD: return madvise_cold(vma, prev, start, end); |
1a4e58cce mm: introduce MAD... |
959 960 |
case MADV_PAGEOUT: return madvise_pageout(vma, prev, start, end); |
854e9ed09 mm: support madvi... |
961 |
case MADV_FREE: |
1da177e4c Linux-2.6.12-rc2 |
962 |
case MADV_DONTNEED: |
230ca982b userfaultfd: non-... |
963 |
return madvise_dontneed_free(vma, prev, start, end, behavior); |
4ca9b3859 mm/madvise: intro... |
964 965 966 |
case MADV_POPULATE_READ: case MADV_POPULATE_WRITE: return madvise_populate(vma, prev, start, end, behavior); |
1da177e4c Linux-2.6.12-rc2 |
967 |
default: |
3866ea90d ksm: first tidy u... |
968 |
return madvise_behavior(vma, prev, start, end, behavior); |
1da177e4c Linux-2.6.12-rc2 |
969 |
} |
1da177e4c Linux-2.6.12-rc2 |
970 |
} |
1ecef9ed0 mm/madvise.c: mak... |
971 |
static bool |
75927af8b mm: madvise(): co... |
972 973 974 975 976 977 978 979 980 981 982 |
madvise_behavior_valid(int behavior) { switch (behavior) { case MADV_DOFORK: case MADV_DONTFORK: case MADV_NORMAL: case MADV_SEQUENTIAL: case MADV_RANDOM: case MADV_REMOVE: case MADV_WILLNEED: case MADV_DONTNEED: |
854e9ed09 mm: support madvi... |
983 |
case MADV_FREE: |
9c276cc65 mm: introduce MAD... |
984 |
case MADV_COLD: |
1a4e58cce mm: introduce MAD... |
985 |
case MADV_PAGEOUT: |
4ca9b3859 mm/madvise: intro... |
986 987 |
case MADV_POPULATE_READ: case MADV_POPULATE_WRITE: |
f8af4da3b ksm: the mm inter... |
988 989 990 991 |
#ifdef CONFIG_KSM case MADV_MERGEABLE: case MADV_UNMERGEABLE: #endif |
0af4e98b6 thp: madvise(MADV... |
992 993 |
#ifdef CONFIG_TRANSPARENT_HUGEPAGE case MADV_HUGEPAGE: |
a664b2d85 thp: madvise(MADV... |
994 |
case MADV_NOHUGEPAGE: |
0af4e98b6 thp: madvise(MADV... |
995 |
#endif |
accb61fe7 coredump: add VM_... |
996 997 |
case MADV_DONTDUMP: case MADV_DODUMP: |
d2cd9ede6 mm,fork: introduc... |
998 999 |
case MADV_WIPEONFORK: case MADV_KEEPONFORK: |
5e451be75 mm/madvise: move ... |
1000 1001 1002 1003 |
#ifdef CONFIG_MEMORY_FAILURE case MADV_SOFT_OFFLINE: case MADV_HWPOISON: #endif |
1ecef9ed0 mm/madvise.c: mak... |
1004 |
return true; |
75927af8b mm: madvise(): co... |
1005 1006 |
default: |
1ecef9ed0 mm/madvise.c: mak... |
1007 |
return false; |
75927af8b mm: madvise(): co... |
1008 1009 |
} } |
3866ea90d ksm: first tidy u... |
1010 |
|
ecb8ac8b1 mm/madvise: intro... |
1011 1012 1013 1014 1015 1016 |
static bool process_madvise_behavior_valid(int behavior) { switch (behavior) { case MADV_COLD: case MADV_PAGEOUT: |
d5fffc5af mm/madvise: add M... |
1017 |
case MADV_WILLNEED: |
ecb8ac8b1 mm/madvise: intro... |
1018 1019 1020 1021 1022 |
return true; default: return false; } } |
1da177e4c Linux-2.6.12-rc2 |
1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 |
/* * The madvise(2) system call. * * Applications can use madvise() to advise the kernel how it should * handle paging I/O in this VM area. The idea is to help the kernel * use appropriate read-ahead and caching techniques. The information * provided is advisory only, and can be safely disregarded by the * kernel without affecting the correct operation of the application. * * behavior values: * MADV_NORMAL - the default behavior is to read clusters. This * results in some read-ahead and read-behind. * MADV_RANDOM - the system should read the minimum amount of data * on any access, since it is unlikely that the appli- * cation will need more than what it asks for. * MADV_SEQUENTIAL - pages in the given range will probably be accessed * once, so they can be aggressively read ahead, and * can be freed soon after they are accessed. * MADV_WILLNEED - the application is notifying the system to read * some pages ahead. * MADV_DONTNEED - the application is finished with the given range, * so the kernel can free resources associated with it. |
d7206a70a mm/madvise: updat... |
1045 1046 |
* MADV_FREE - the application marks pages in the given range as lazy free, * where actual purges are postponed until memory pressure happens. |
f6b3ec238 [PATCH] madvise(M... |
1047 1048 |
* MADV_REMOVE - the application wants to free up the given range of * pages and associated backing store. |
3866ea90d ksm: first tidy u... |
1049 1050 1051 |
* MADV_DONTFORK - omit this area from child's address space when forking: * typically, to avoid COWing pages pinned by get_user_pages(). * MADV_DOFORK - cancel MADV_DONTFORK: no longer omit this area when forking. |
c02c30093 mm/madvise.c: add... |
1052 1053 1054 |
* MADV_WIPEONFORK - present the child process with zero-filled memory in this * range after a fork. * MADV_KEEPONFORK - undo the effect of MADV_WIPEONFORK |
d7206a70a mm/madvise: updat... |
1055 1056 1057 |
* MADV_HWPOISON - trigger memory error handler as if the given memory range * were corrupted by unrecoverable hardware memory failure. * MADV_SOFT_OFFLINE - try to soft-offline the given range of memory. |
f8af4da3b ksm: the mm inter... |
1058 1059 1060 |
* MADV_MERGEABLE - the application recommends that KSM try to merge pages in * this area with pages of identical content from other such areas. * MADV_UNMERGEABLE- cancel MADV_MERGEABLE: no longer merge pages with others. |
d7206a70a mm/madvise: updat... |
1061 1062 1063 1064 1065 1066 1067 1068 1069 |
* MADV_HUGEPAGE - the application wants to back the given range by transparent * huge pages in the future. Existing pages might be coalesced and * new pages might be allocated as THP. * MADV_NOHUGEPAGE - mark the given range as not worth being backed by * transparent huge pages so the existing pages will not be * coalesced into THP and new pages will not be allocated as THP. * MADV_DONTDUMP - the application wants to prevent pages in the given range * from being included in its core dump. * MADV_DODUMP - cancel MADV_DONTDUMP: no longer exclude from core dump. |
ecb8ac8b1 mm/madvise: intro... |
1070 1071 |
* MADV_COLD - the application is not expected to use this memory soon, * deactivate pages in this range so that they can be reclaimed |
f0953a1bb mm: fix typos in ... |
1072 |
* easily if memory pressure happens. |
ecb8ac8b1 mm/madvise: intro... |
1073 1074 |
* MADV_PAGEOUT - the application is not expected to use this memory soon, * page out the pages in this range immediately. |
4ca9b3859 mm/madvise: intro... |
1075 1076 1077 1078 |
* MADV_POPULATE_READ - populate (prefault) page tables readable by * triggering read faults if required * MADV_POPULATE_WRITE - populate (prefault) page tables writable by * triggering write faults if required |
1da177e4c Linux-2.6.12-rc2 |
1079 1080 1081 1082 1083 |
* * return values: * zero - success * -EINVAL - start + len < 0, start is not page-aligned, * "behavior" is not a valid value, or application |
c02c30093 mm/madvise.c: add... |
1084 1085 1086 |
* is attempting to release locked or shared pages, * or the specified address range includes file, Huge TLB, * MAP_SHARED or VMPFNMAP range. |
1da177e4c Linux-2.6.12-rc2 |
1087 1088 1089 1090 1091 1092 |
* -ENOMEM - addresses in the specified range are not currently * mapped, or are outside the AS of the process. * -EIO - an I/O error occurred while paging in data. * -EBADF - map exists, but area maps something that isn't a file. * -EAGAIN - a kernel resource was temporarily unavailable. */ |
0726b01e7 mm/madvise: pass ... |
1093 |
int do_madvise(struct mm_struct *mm, unsigned long start, size_t len_in, int behavior) |
1da177e4c Linux-2.6.12-rc2 |
1094 |
{ |
05b743847 [PATCH] madvise: ... |
1095 |
unsigned long end, tmp; |
ec9bed9d3 mm/madvise.c: fix... |
1096 |
struct vm_area_struct *vma, *prev; |
1da177e4c Linux-2.6.12-rc2 |
1097 1098 |
int unmapped_error = 0; int error = -EINVAL; |
f79777932 speed up madvise_... |
1099 |
int write; |
1da177e4c Linux-2.6.12-rc2 |
1100 |
size_t len; |
1998cc048 mm: make madvise(... |
1101 |
struct blk_plug plug; |
1da177e4c Linux-2.6.12-rc2 |
1102 |
|
057d33891 mm: untag user po... |
1103 |
start = untagged_addr(start); |
75927af8b mm: madvise(): co... |
1104 1105 |
if (!madvise_behavior_valid(behavior)) return error; |
df6c6500b mm/madvise.c: use... |
1106 |
if (!PAGE_ALIGNED(start)) |
84d96d897 mm: madvise: comp... |
1107 |
return error; |
df6c6500b mm/madvise.c: use... |
1108 |
len = PAGE_ALIGN(len_in); |
1da177e4c Linux-2.6.12-rc2 |
1109 1110 1111 |
/* Check to see whether len was rounded up from small -ve to zero */ if (len_in && !len) |
84d96d897 mm: madvise: comp... |
1112 |
return error; |
1da177e4c Linux-2.6.12-rc2 |
1113 1114 1115 |
end = start + len; if (end < start) |
84d96d897 mm: madvise: comp... |
1116 |
return error; |
1da177e4c Linux-2.6.12-rc2 |
1117 1118 1119 |
error = 0; if (end == start) |
84d96d897 mm: madvise: comp... |
1120 |
return error; |
5e451be75 mm/madvise: move ... |
1121 1122 1123 1124 |
#ifdef CONFIG_MEMORY_FAILURE if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE) return madvise_inject_error(behavior, start, start + len_in); #endif |
84d96d897 mm: madvise: comp... |
1125 |
write = madvise_need_mmap_write(behavior); |
dc0ef0df7 mm: make mmap_sem... |
1126 |
if (write) { |
0726b01e7 mm/madvise: pass ... |
1127 |
if (mmap_write_lock_killable(mm)) |
dc0ef0df7 mm: make mmap_sem... |
1128 1129 |
return -EINTR; } else { |
0726b01e7 mm/madvise: pass ... |
1130 |
mmap_read_lock(mm); |
dc0ef0df7 mm: make mmap_sem... |
1131 |
} |
1da177e4c Linux-2.6.12-rc2 |
1132 1133 1134 1135 |
/* * If the interval [start,end) covers some unmapped address * ranges, just ignore them, but return -ENOMEM at the end. |
05b743847 [PATCH] madvise: ... |
1136 |
* - different from the way of handling in mlock etc. |
1da177e4c Linux-2.6.12-rc2 |
1137 |
*/ |
0726b01e7 mm/madvise: pass ... |
1138 |
vma = find_vma_prev(mm, start, &prev); |
836d5ffd3 [PATCH] mm: fix m... |
1139 1140 |
if (vma && start > vma->vm_start) prev = vma; |
1998cc048 mm: make madvise(... |
1141 |
blk_start_plug(&plug); |
1da177e4c Linux-2.6.12-rc2 |
1142 1143 1144 1145 |
for (;;) { /* Still start < end. */ error = -ENOMEM; if (!vma) |
84d96d897 mm: madvise: comp... |
1146 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
1147 |
|
05b743847 [PATCH] madvise: ... |
1148 |
/* Here start < (end|vma->vm_end). */ |
1da177e4c Linux-2.6.12-rc2 |
1149 1150 1151 |
if (start < vma->vm_start) { unmapped_error = -ENOMEM; start = vma->vm_start; |
05b743847 [PATCH] madvise: ... |
1152 |
if (start >= end) |
84d96d897 mm: madvise: comp... |
1153 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
1154 |
} |
05b743847 [PATCH] madvise: ... |
1155 1156 1157 1158 |
/* Here vma->vm_start <= start < (end|vma->vm_end) */ tmp = vma->vm_end; if (end < tmp) tmp = end; |
1da177e4c Linux-2.6.12-rc2 |
1159 |
|
05b743847 [PATCH] madvise: ... |
1160 1161 |
/* Here vma->vm_start <= start < tmp <= (end|vma->vm_end). */ error = madvise_vma(vma, &prev, start, tmp, behavior); |
1da177e4c Linux-2.6.12-rc2 |
1162 |
if (error) |
84d96d897 mm: madvise: comp... |
1163 |
goto out; |
05b743847 [PATCH] madvise: ... |
1164 |
start = tmp; |
90ed52ebe [PATCH] holepunch... |
1165 |
if (prev && start < prev->vm_end) |
05b743847 [PATCH] madvise: ... |
1166 1167 1168 |
start = prev->vm_end; error = unmapped_error; if (start >= end) |
84d96d897 mm: madvise: comp... |
1169 |
goto out; |
90ed52ebe [PATCH] holepunch... |
1170 1171 |
if (prev) vma = prev->vm_next; |
c1e8d7c6a mmap locking API:... |
1172 |
else /* madvise_remove dropped mmap_lock */ |
0726b01e7 mm/madvise: pass ... |
1173 |
vma = find_vma(mm, start); |
1da177e4c Linux-2.6.12-rc2 |
1174 |
} |
1da177e4c Linux-2.6.12-rc2 |
1175 |
out: |
84d96d897 mm: madvise: comp... |
1176 |
blk_finish_plug(&plug); |
f79777932 speed up madvise_... |
1177 |
if (write) |
0726b01e7 mm/madvise: pass ... |
1178 |
mmap_write_unlock(mm); |
0a27a14a6 mm: madvise avoid... |
1179 |
else |
0726b01e7 mm/madvise: pass ... |
1180 |
mmap_read_unlock(mm); |
0a27a14a6 mm: madvise avoid... |
1181 |
|
1da177e4c Linux-2.6.12-rc2 |
1182 1183 |
return error; } |
db08ca252 mm: make do_madvi... |
1184 1185 1186 |
SYSCALL_DEFINE3(madvise, unsigned long, start, size_t, len_in, int, behavior) { |
0726b01e7 mm/madvise: pass ... |
1187 |
return do_madvise(current->mm, start, len_in, behavior); |
db08ca252 mm: make do_madvi... |
1188 |
} |
ecb8ac8b1 mm/madvise: intro... |
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 1214 1215 1216 1217 1218 1219 1220 1221 1222 |
SYSCALL_DEFINE5(process_madvise, int, pidfd, const struct iovec __user *, vec, size_t, vlen, int, behavior, unsigned int, flags) { ssize_t ret; struct iovec iovstack[UIO_FASTIOV], iovec; struct iovec *iov = iovstack; struct iov_iter iter; struct pid *pid; struct task_struct *task; struct mm_struct *mm; size_t total_len; unsigned int f_flags; if (flags != 0) { ret = -EINVAL; goto out; } ret = import_iovec(READ, vec, vlen, ARRAY_SIZE(iovstack), &iov, &iter); if (ret < 0) goto out; pid = pidfd_get_pid(pidfd, &f_flags); if (IS_ERR(pid)) { ret = PTR_ERR(pid); goto free_iov; } task = get_pid_task(pid, PIDTYPE_PID); if (!task) { ret = -ESRCH; goto put_pid; } |
a68a0262a mm/madvise: remov... |
1223 |
if (!process_madvise_behavior_valid(behavior)) { |
ecb8ac8b1 mm/madvise: intro... |
1224 1225 1226 |
ret = -EINVAL; goto release_task; } |
96cfe2c0f mm/madvise: repla... |
1227 1228 |
/* Require PTRACE_MODE_READ to avoid leaking ASLR metadata. */ mm = mm_access(task, PTRACE_MODE_READ_FSCREDS); |
ecb8ac8b1 mm/madvise: intro... |
1229 1230 1231 1232 |
if (IS_ERR_OR_NULL(mm)) { ret = IS_ERR(mm) ? PTR_ERR(mm) : -ESRCH; goto release_task; } |
96cfe2c0f mm/madvise: repla... |
1233 1234 1235 1236 1237 1238 1239 1240 |
/* * Require CAP_SYS_NICE for influencing process performance. Note that * only non-destructive hints are currently supported. */ if (!capable(CAP_SYS_NICE)) { ret = -EPERM; goto release_mm; } |
ecb8ac8b1 mm/madvise: intro... |
1241 1242 1243 1244 1245 1246 |
total_len = iov_iter_count(&iter); while (iov_iter_count(&iter)) { iovec = iov_iter_iovec(&iter); ret = do_madvise(mm, (unsigned long)iovec.iov_base, iovec.iov_len, behavior); |
d4835551f Revert "mm: madvi... |
1247 |
if (ret < 0) |
ecb8ac8b1 mm/madvise: intro... |
1248 1249 1250 |
break; iov_iter_advance(&iter, iovec.iov_len); } |
a07a4b75c mm: madvise: retu... |
1251 |
ret = (total_len - iov_iter_count(&iter)) ? : ret; |
ecb8ac8b1 mm/madvise: intro... |
1252 |
|
96cfe2c0f mm/madvise: repla... |
1253 |
release_mm: |
ecb8ac8b1 mm/madvise: intro... |
1254 |
mmput(mm); |
ecb8ac8b1 mm/madvise: intro... |
1255 1256 1257 1258 1259 1260 1261 1262 1263 |
release_task: put_task_struct(task); put_pid: put_pid(pid); free_iov: kfree(iov); out: return ret; } |