Commit | Line | Data |
---|---|---|
b2441318 | 1 | /* SPDX-License-Identifier: GPL-2.0 */ |
1da177e4 LT |
2 | /* |
3 | * Macros for manipulating and testing page->flags | |
4 | */ | |
5 | ||
6 | #ifndef PAGE_FLAGS_H | |
7 | #define PAGE_FLAGS_H | |
8 | ||
f886ed44 | 9 | #include <linux/types.h> |
187f1882 | 10 | #include <linux/bug.h> |
072bb0aa | 11 | #include <linux/mmdebug.h> |
9223b419 | 12 | #ifndef __GENERATING_BOUNDS_H |
6d777953 | 13 | #include <linux/mm_types.h> |
01fc0ac1 | 14 | #include <generated/bounds.h> |
9223b419 | 15 | #endif /* !__GENERATING_BOUNDS_H */ |
f886ed44 | 16 | |
1da177e4 LT |
17 | /* |
18 | * Various page->flags bits: | |
19 | * | |
6e2e07cd DH |
20 | * PG_reserved is set for special pages. The "struct page" of such a page |
21 | * should in general not be touched (e.g. set dirty) except by its owner. | |
22 | * Pages marked as PG_reserved include: | |
23 | * - Pages part of the kernel image (including vDSO) and similar (e.g. BIOS, | |
24 | * initrd, HW tables) | |
25 | * - Pages reserved or allocated early during boot (before the page allocator | |
26 | * was initialized). This includes (depending on the architecture) the | |
27 | * initial vmemmap, initial page tables, crashkernel, elfcorehdr, and much | |
28 | * much more. Once (if ever) freed, PG_reserved is cleared and they will | |
29 | * be given to the page allocator. | |
30 | * - Pages falling into physical memory gaps - not IORESOURCE_SYSRAM. Trying | |
31 | * to read/write these pages might end badly. Don't touch! | |
32 | * - The zero page(s) | |
6e2e07cd DH |
33 | * - Pages allocated in the context of kexec/kdump (loaded kernel image, |
34 | * control pages, vmcoreinfo) | |
35 | * - MMIO/DMA pages. Some architectures don't allow to ioremap pages that are | |
36 | * not marked PG_reserved (as they might be in use by somebody else who does | |
37 | * not respect the caching strategy). | |
6e2e07cd DH |
38 | * - MCA pages on ia64 |
39 | * - Pages holding CPU notes for POWER Firmware Assisted Dump | |
40 | * - Device memory (e.g. PMEM, DAX, HMM) | |
41 | * Some PG_reserved pages will be excluded from the hibernation image. | |
42 | * PG_reserved does in general not hinder anybody from dumping or swapping | |
43 | * and is no longer required for remap_pfn_range(). ioremap might require it. | |
44 | * Consequently, PG_reserved for a page mapped into user space can indicate | |
45 | * the zero page, the vDSO, MMIO pages or device memory. | |
1da177e4 | 46 | * |
da6052f7 NP |
47 | * The PG_private bitflag is set on pagecache pages if they contain filesystem |
48 | * specific data (which is normally at page->private). It can be used by | |
49 | * private allocations for its own usage. | |
1da177e4 | 50 | * |
da6052f7 NP |
51 | * During initiation of disk I/O, PG_locked is set. This bit is set before I/O |
52 | * and cleared when writeback _starts_ or when read _completes_. PG_writeback | |
53 | * is set before writeback starts and cleared when it finishes. | |
54 | * | |
55 | * PG_locked also pins a page in pagecache, and blocks truncation of the file | |
56 | * while it is held. | |
57 | * | |
58 | * page_waitqueue(page) is a wait queue of all tasks waiting for the page | |
59 | * to become unlocked. | |
1da177e4 | 60 | * |
9de4f22a HY |
61 | * PG_swapbacked is set when a page uses swap as a backing storage. This are |
62 | * usually PageAnon or shmem pages but please note that even anonymous pages | |
63 | * might lose their PG_swapbacked flag when they simply can be dropped (e.g. as | |
64 | * a result of MADV_FREE). | |
65 | * | |
da6052f7 NP |
66 | * PG_referenced, PG_reclaim are used for page reclaim for anonymous and |
67 | * file-backed pagecache (see mm/vmscan.c). | |
1da177e4 | 68 | * |
1da177e4 LT |
69 | * PG_arch_1 is an architecture specific page state bit. The generic code |
70 | * guarantees that this bit is cleared for a page when it first is entered into | |
71 | * the page cache. | |
72 | * | |
d466f2fc AK |
73 | * PG_hwpoison indicates that a page got corrupted in hardware and contains |
74 | * data with incorrect ECC bits that triggered a machine check. Accessing is | |
75 | * not safe since it may cause another machine check. Don't touch! | |
1da177e4 LT |
76 | */ |
77 | ||
78 | /* | |
3b12da6d | 79 | * Don't use the pageflags directly. Use the PageFoo macros. |
91fc8ab3 AW |
80 | * |
81 | * The page flags field is split into two parts, the main flags area | |
82 | * which extends from the low bits upwards, and the fields area which | |
83 | * extends from the high bits downwards. | |
84 | * | |
85 | * | FIELD | ... | FLAGS | | |
9223b419 CL |
86 | * N-1 ^ 0 |
87 | * (NR_PAGEFLAGS) | |
91fc8ab3 | 88 | * |
9223b419 CL |
89 | * The fields area is reserved for fields mapping zone, node (for NUMA) and |
90 | * SPARSEMEM section (for variants of SPARSEMEM that require section ids like | |
91 | * SPARSEMEM_EXTREME with !SPARSEMEM_VMEMMAP). | |
1da177e4 | 92 | */ |
e2683181 CL |
93 | enum pageflags { |
94 | PG_locked, /* Page is locked. Don't touch. */ | |
c704ae97 | 95 | PG_writeback, /* Page is under writeback */ |
e2683181 CL |
96 | PG_referenced, |
97 | PG_uptodate, | |
98 | PG_dirty, | |
99 | PG_lru, | |
c704ae97 MWO |
100 | PG_head, /* Must be in bit 6 */ |
101 | PG_waiters, /* Page has waiters, check its waitqueue. Must be bit #7 and in the same byte as "PG_locked" */ | |
e2683181 | 102 | PG_active, |
1899ad18 | 103 | PG_workingset, |
02e1960a MWO |
104 | PG_owner_priv_1, /* Owner use. If pagecache, fs may use */ |
105 | PG_owner_2, /* Owner use. If pagecache, fs may use */ | |
e2683181 CL |
106 | PG_arch_1, |
107 | PG_reserved, | |
108 | PG_private, /* If pagecache, has fs-private data */ | |
266cf658 | 109 | PG_private_2, /* If pagecache, has fs aux data */ |
e2683181 | 110 | PG_reclaim, /* To be reclaimed asap */ |
b2e18538 | 111 | PG_swapbacked, /* Page is backed by RAM/swap */ |
894bc310 | 112 | PG_unevictable, /* Page is "unevictable" */ |
af8e3354 | 113 | #ifdef CONFIG_MMU |
b291f000 | 114 | PG_mlocked, /* Page is vma mlocked */ |
894bc310 | 115 | #endif |
d466f2fc AK |
116 | #ifdef CONFIG_MEMORY_FAILURE |
117 | PG_hwpoison, /* hardware poisoned page. Don't touch */ | |
e9da73d6 | 118 | #endif |
1c676e0d | 119 | #if defined(CONFIG_PAGE_IDLE_FLAG) && defined(CONFIG_64BIT) |
33c3fc71 VD |
120 | PG_young, |
121 | PG_idle, | |
4beba948 | 122 | #endif |
7a87225a | 123 | #ifdef CONFIG_ARCH_USES_PG_ARCH_2 |
4beba948 | 124 | PG_arch_2, |
7a87225a MWO |
125 | #endif |
126 | #ifdef CONFIG_ARCH_USES_PG_ARCH_3 | |
ef6458b1 | 127 | PG_arch_3, |
f886ed44 | 128 | #endif |
0cad47cf AW |
129 | __NR_PAGEFLAGS, |
130 | ||
d389a4a8 MWO |
131 | PG_readahead = PG_reclaim, |
132 | ||
02e1960a MWO |
133 | /* Anonymous memory (and shmem) */ |
134 | PG_swapcache = PG_owner_priv_1, /* Swap page: swp_entry_t in private */ | |
135 | /* Some filesystems */ | |
136 | PG_checked = PG_owner_priv_1, | |
137 | ||
78fbe906 DH |
138 | /* |
139 | * Depending on the way an anonymous folio can be mapped into a page | |
140 | * table (e.g., single PMD/PUD/CONT of the head page vs. PTE-mapped | |
141 | * THP), PG_anon_exclusive may be set only for the head page or for | |
142 | * tail pages of an anonymous folio. For now, we only expect it to be | |
143 | * set on tail pages for PTE-mapped THP. | |
144 | */ | |
02e1960a | 145 | PG_anon_exclusive = PG_owner_2, |
0cad47cf | 146 | |
02e1960a MWO |
147 | /* |
148 | * Set if all buffer heads in the folio are mapped. | |
149 | * Filesystems which do not use BHs can use it for their own purpose. | |
150 | */ | |
151 | PG_mappedtodisk = PG_owner_2, | |
6326fec1 | 152 | |
266cf658 DH |
153 | /* Two page bits are conscripted by FS-Cache to maintain local caching |
154 | * state. These bits are set on pages belonging to the netfs's inodes | |
155 | * when those inodes are being locally cached. | |
156 | */ | |
157 | PG_fscache = PG_private_2, /* page backed by cache */ | |
158 | ||
0cad47cf | 159 | /* XEN */ |
d8ac3dd4 | 160 | /* Pinned in Xen as a read-only pagetable page. */ |
0cad47cf | 161 | PG_pinned = PG_owner_priv_1, |
d8ac3dd4 | 162 | /* Pinned as part of domain save (see xen_mm_pin_all()). */ |
0cad47cf | 163 | PG_savepinned = PG_dirty, |
d8ac3dd4 JH |
164 | /* Has a grant mapping of another (foreign) domain's page. */ |
165 | PG_foreign = PG_owner_priv_1, | |
b877ac98 JG |
166 | /* Remapped by swiotlb-xen. */ |
167 | PG_xen_remapped = PG_owner_priv_1, | |
8a38082d | 168 | |
bda807d4 MK |
169 | /* non-lru isolated movable page */ |
170 | PG_isolated = PG_reclaim, | |
36e66c55 AD |
171 | |
172 | /* Only valid for buddy pages. Used to track pages that are reported */ | |
173 | PG_reported = PG_uptodate, | |
66361095 MS |
174 | |
175 | #ifdef CONFIG_MEMORY_HOTPLUG | |
176 | /* For self-hosted memmap pages */ | |
177 | PG_vmemmap_self_hosted = PG_owner_priv_1, | |
178 | #endif | |
9c5ccf2d MWO |
179 | |
180 | /* | |
181 | * Flags only valid for compound pages. Stored in first tail page's | |
ebc1baf5 MWO |
182 | * flags word. Cannot use the first 8 flags or any flag marked as |
183 | * PF_ANY. | |
9c5ccf2d MWO |
184 | */ |
185 | ||
186 | /* At least one page in this folio has the hwpoison flag set */ | |
09022bc1 | 187 | PG_has_hwpoisoned = PG_active, |
de53c05f | 188 | PG_large_rmappable = PG_workingset, /* anon or file-backed */ |
8422acdc | 189 | PG_partially_mapped = PG_reclaim, /* was identified to be partially mapped */ |
e2683181 | 190 | }; |
1da177e4 | 191 | |
41c961b9 MS |
192 | #define PAGEFLAGS_MASK ((1UL << NR_PAGEFLAGS) - 1) |
193 | ||
9223b419 CL |
194 | #ifndef __GENERATING_BOUNDS_H |
195 | ||
47010c04 | 196 | #ifdef CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP |
cf5472e5 | 197 | DECLARE_STATIC_KEY_FALSE(hugetlb_optimize_vmemmap_key); |
a6b40850 | 198 | |
e7d32485 | 199 | /* |
838691a1 MS |
200 | * Return the real head page struct iff the @page is a fake head page, otherwise |
201 | * return the @page itself. See Documentation/mm/vmemmap_dedup.rst. | |
e7d32485 MS |
202 | */ |
203 | static __always_inline const struct page *page_fixed_fake_head(const struct page *page) | |
204 | { | |
cf5472e5 | 205 | if (!static_branch_unlikely(&hugetlb_optimize_vmemmap_key)) |
e7d32485 MS |
206 | return page; |
207 | ||
208 | /* | |
209 | * Only addresses aligned with PAGE_SIZE of struct page may be fake head | |
210 | * struct page. The alignment check aims to avoid access the fields ( | |
211 | * e.g. compound_head) of the @page[1]. It can avoid touch a (possibly) | |
212 | * cold cacheline in some cases. | |
213 | */ | |
214 | if (IS_ALIGNED((unsigned long)page, PAGE_SIZE) && | |
215 | test_bit(PG_head, &page->flags)) { | |
216 | /* | |
217 | * We can safely access the field of the @page[1] with PG_head | |
218 | * because the @page is a compound page composed with at least | |
219 | * two contiguous pages. | |
220 | */ | |
221 | unsigned long head = READ_ONCE(page[1].compound_head); | |
222 | ||
223 | if (likely(head & 1)) | |
224 | return (const struct page *)(head - 1); | |
225 | } | |
226 | return page; | |
227 | } | |
228 | #else | |
229 | static inline const struct page *page_fixed_fake_head(const struct page *page) | |
230 | { | |
231 | return page; | |
232 | } | |
233 | #endif | |
234 | ||
ce3467af | 235 | static __always_inline int page_is_fake_head(const struct page *page) |
e7d32485 MS |
236 | { |
237 | return page_fixed_fake_head(page) != page; | |
238 | } | |
239 | ||
ef5f379d | 240 | static __always_inline unsigned long _compound_head(const struct page *page) |
0e6d31a7 KS |
241 | { |
242 | unsigned long head = READ_ONCE(page->compound_head); | |
243 | ||
244 | if (unlikely(head & 1)) | |
0f2317e3 | 245 | return head - 1; |
e7d32485 | 246 | return (unsigned long)page_fixed_fake_head(page); |
0e6d31a7 KS |
247 | } |
248 | ||
0f2317e3 MWO |
249 | #define compound_head(page) ((typeof(page))_compound_head(page)) |
250 | ||
7b230db3 MWO |
251 | /** |
252 | * page_folio - Converts from page to folio. | |
253 | * @p: The page. | |
254 | * | |
255 | * Every page is part of a folio. This function cannot be called on a | |
256 | * NULL pointer. | |
257 | * | |
258 | * Context: No reference, nor lock is required on @page. If the caller | |
259 | * does not hold a reference, this call may race with a folio split, so | |
260 | * it should re-check the folio still contains this page after gaining | |
261 | * a reference on the folio. | |
262 | * Return: The folio which contains this page. | |
263 | */ | |
264 | #define page_folio(p) (_Generic((p), \ | |
265 | const struct page *: (const struct folio *)_compound_head(p), \ | |
266 | struct page *: (struct folio *)_compound_head(p))) | |
267 | ||
268 | /** | |
269 | * folio_page - Return a page from a folio. | |
270 | * @folio: The folio. | |
271 | * @n: The page number to return. | |
272 | * | |
273 | * @n is relative to the start of the folio. This function does not | |
274 | * check that the page number lies within @folio; the caller is presumed | |
275 | * to have a reference to the page. | |
276 | */ | |
277 | #define folio_page(folio, n) nth_page(&(folio)->page, n) | |
278 | ||
ce3467af | 279 | static __always_inline int PageTail(const struct page *page) |
0e6d31a7 | 280 | { |
e7d32485 | 281 | return READ_ONCE(page->compound_head) & 1 || page_is_fake_head(page); |
0e6d31a7 KS |
282 | } |
283 | ||
ce3467af | 284 | static __always_inline int PageCompound(const struct page *page) |
0e6d31a7 | 285 | { |
e7d32485 MS |
286 | return test_bit(PG_head, &page->flags) || |
287 | READ_ONCE(page->compound_head) & 1; | |
0e6d31a7 KS |
288 | } |
289 | ||
f165b378 PT |
290 | #define PAGE_POISON_PATTERN -1l |
291 | static inline int PagePoisoned(const struct page *page) | |
292 | { | |
477d01fc | 293 | return READ_ONCE(page->flags) == PAGE_POISON_PATTERN; |
f165b378 PT |
294 | } |
295 | ||
f682a97a AD |
296 | #ifdef CONFIG_DEBUG_VM |
297 | void page_init_poison(struct page *page, size_t size); | |
298 | #else | |
299 | static inline void page_init_poison(struct page *page, size_t size) | |
300 | { | |
301 | } | |
302 | #endif | |
303 | ||
ce3467af MWO |
304 | static const unsigned long *const_folio_flags(const struct folio *folio, |
305 | unsigned n) | |
306 | { | |
307 | const struct page *page = &folio->page; | |
308 | ||
309 | VM_BUG_ON_PGFLAGS(PageTail(page), page); | |
310 | VM_BUG_ON_PGFLAGS(n > 0 && !test_bit(PG_head, &page->flags), page); | |
311 | return &page[n].flags; | |
312 | } | |
313 | ||
d389a4a8 MWO |
314 | static unsigned long *folio_flags(struct folio *folio, unsigned n) |
315 | { | |
316 | struct page *page = &folio->page; | |
317 | ||
318 | VM_BUG_ON_PGFLAGS(PageTail(page), page); | |
319 | VM_BUG_ON_PGFLAGS(n > 0 && !test_bit(PG_head, &page->flags), page); | |
320 | return &page[n].flags; | |
321 | } | |
322 | ||
95ad9755 KS |
323 | /* |
324 | * Page flags policies wrt compound pages | |
325 | * | |
f165b378 PT |
326 | * PF_POISONED_CHECK |
327 | * check if this struct page poisoned/uninitialized | |
328 | * | |
95ad9755 KS |
329 | * PF_ANY: |
330 | * the page flag is relevant for small, head and tail pages. | |
331 | * | |
332 | * PF_HEAD: | |
333 | * for compound page all operations related to the page flag applied to | |
334 | * head page. | |
335 | * | |
336 | * PF_NO_TAIL: | |
337 | * modifications of the page flag must be done on small or head pages, | |
338 | * checks can be done on tail pages too. | |
339 | * | |
340 | * PF_NO_COMPOUND: | |
341 | * the page flag is not relevant for compound pages. | |
a08d93e5 MWO |
342 | * |
343 | * PF_SECOND: | |
344 | * the page flag is stored in the first tail page. | |
95ad9755 | 345 | */ |
f165b378 PT |
346 | #define PF_POISONED_CHECK(page) ({ \ |
347 | VM_BUG_ON_PGFLAGS(PagePoisoned(page), page); \ | |
348 | page; }) | |
349 | #define PF_ANY(page, enforce) PF_POISONED_CHECK(page) | |
350 | #define PF_HEAD(page, enforce) PF_POISONED_CHECK(compound_head(page)) | |
95ad9755 KS |
351 | #define PF_NO_TAIL(page, enforce) ({ \ |
352 | VM_BUG_ON_PGFLAGS(enforce && PageTail(page), page); \ | |
f165b378 | 353 | PF_POISONED_CHECK(compound_head(page)); }) |
822cdd11 | 354 | #define PF_NO_COMPOUND(page, enforce) ({ \ |
95ad9755 | 355 | VM_BUG_ON_PGFLAGS(enforce && PageCompound(page), page); \ |
f165b378 | 356 | PF_POISONED_CHECK(page); }) |
a08d93e5 MWO |
357 | #define PF_SECOND(page, enforce) ({ \ |
358 | VM_BUG_ON_PGFLAGS(!PageHead(page), page); \ | |
359 | PF_POISONED_CHECK(&page[1]); }) | |
95ad9755 | 360 | |
d389a4a8 MWO |
361 | /* Which page is the flag stored in */ |
362 | #define FOLIO_PF_ANY 0 | |
363 | #define FOLIO_PF_HEAD 0 | |
d389a4a8 MWO |
364 | #define FOLIO_PF_NO_TAIL 0 |
365 | #define FOLIO_PF_NO_COMPOUND 0 | |
366 | #define FOLIO_PF_SECOND 1 | |
367 | ||
dfbac6dc MWO |
368 | #define FOLIO_HEAD_PAGE 0 |
369 | #define FOLIO_SECOND_PAGE 1 | |
370 | ||
f94a62e9 CL |
371 | /* |
372 | * Macros to create function definitions for page flags | |
373 | */ | |
dfbac6dc | 374 | #define FOLIO_TEST_FLAG(name, page) \ |
ce3467af MWO |
375 | static __always_inline bool folio_test_##name(const struct folio *folio) \ |
376 | { return test_bit(PG_##name, const_folio_flags(folio, page)); } | |
dfbac6dc MWO |
377 | |
378 | #define FOLIO_SET_FLAG(name, page) \ | |
379 | static __always_inline void folio_set_##name(struct folio *folio) \ | |
380 | { set_bit(PG_##name, folio_flags(folio, page)); } | |
381 | ||
382 | #define FOLIO_CLEAR_FLAG(name, page) \ | |
383 | static __always_inline void folio_clear_##name(struct folio *folio) \ | |
384 | { clear_bit(PG_##name, folio_flags(folio, page)); } | |
385 | ||
386 | #define __FOLIO_SET_FLAG(name, page) \ | |
387 | static __always_inline void __folio_set_##name(struct folio *folio) \ | |
388 | { __set_bit(PG_##name, folio_flags(folio, page)); } | |
389 | ||
390 | #define __FOLIO_CLEAR_FLAG(name, page) \ | |
391 | static __always_inline void __folio_clear_##name(struct folio *folio) \ | |
392 | { __clear_bit(PG_##name, folio_flags(folio, page)); } | |
393 | ||
394 | #define FOLIO_TEST_SET_FLAG(name, page) \ | |
395 | static __always_inline bool folio_test_set_##name(struct folio *folio) \ | |
396 | { return test_and_set_bit(PG_##name, folio_flags(folio, page)); } | |
397 | ||
398 | #define FOLIO_TEST_CLEAR_FLAG(name, page) \ | |
399 | static __always_inline bool folio_test_clear_##name(struct folio *folio) \ | |
400 | { return test_and_clear_bit(PG_##name, folio_flags(folio, page)); } | |
401 | ||
402 | #define FOLIO_FLAG(name, page) \ | |
403 | FOLIO_TEST_FLAG(name, page) \ | |
404 | FOLIO_SET_FLAG(name, page) \ | |
405 | FOLIO_CLEAR_FLAG(name, page) | |
406 | ||
95ad9755 | 407 | #define TESTPAGEFLAG(uname, lname, policy) \ |
dfbac6dc | 408 | FOLIO_TEST_FLAG(lname, FOLIO_##policy) \ |
ce3467af | 409 | static __always_inline int Page##uname(const struct page *page) \ |
d389a4a8 | 410 | { return test_bit(PG_##lname, &policy(page, 0)->flags); } |
f94a62e9 | 411 | |
95ad9755 | 412 | #define SETPAGEFLAG(uname, lname, policy) \ |
dfbac6dc | 413 | FOLIO_SET_FLAG(lname, FOLIO_##policy) \ |
4b0f3261 | 414 | static __always_inline void SetPage##uname(struct page *page) \ |
d389a4a8 | 415 | { set_bit(PG_##lname, &policy(page, 1)->flags); } |
f94a62e9 | 416 | |
95ad9755 | 417 | #define CLEARPAGEFLAG(uname, lname, policy) \ |
dfbac6dc | 418 | FOLIO_CLEAR_FLAG(lname, FOLIO_##policy) \ |
4b0f3261 | 419 | static __always_inline void ClearPage##uname(struct page *page) \ |
d389a4a8 | 420 | { clear_bit(PG_##lname, &policy(page, 1)->flags); } |
f94a62e9 | 421 | |
95ad9755 | 422 | #define __SETPAGEFLAG(uname, lname, policy) \ |
dfbac6dc | 423 | __FOLIO_SET_FLAG(lname, FOLIO_##policy) \ |
4b0f3261 | 424 | static __always_inline void __SetPage##uname(struct page *page) \ |
d389a4a8 | 425 | { __set_bit(PG_##lname, &policy(page, 1)->flags); } |
f94a62e9 | 426 | |
95ad9755 | 427 | #define __CLEARPAGEFLAG(uname, lname, policy) \ |
dfbac6dc | 428 | __FOLIO_CLEAR_FLAG(lname, FOLIO_##policy) \ |
4b0f3261 | 429 | static __always_inline void __ClearPage##uname(struct page *page) \ |
d389a4a8 | 430 | { __clear_bit(PG_##lname, &policy(page, 1)->flags); } |
f94a62e9 | 431 | |
95ad9755 | 432 | #define TESTSETFLAG(uname, lname, policy) \ |
dfbac6dc | 433 | FOLIO_TEST_SET_FLAG(lname, FOLIO_##policy) \ |
4b0f3261 | 434 | static __always_inline int TestSetPage##uname(struct page *page) \ |
d389a4a8 | 435 | { return test_and_set_bit(PG_##lname, &policy(page, 1)->flags); } |
f94a62e9 | 436 | |
95ad9755 | 437 | #define TESTCLEARFLAG(uname, lname, policy) \ |
dfbac6dc | 438 | FOLIO_TEST_CLEAR_FLAG(lname, FOLIO_##policy) \ |
4b0f3261 | 439 | static __always_inline int TestClearPage##uname(struct page *page) \ |
d389a4a8 | 440 | { return test_and_clear_bit(PG_##lname, &policy(page, 1)->flags); } |
f94a62e9 | 441 | |
95ad9755 KS |
442 | #define PAGEFLAG(uname, lname, policy) \ |
443 | TESTPAGEFLAG(uname, lname, policy) \ | |
444 | SETPAGEFLAG(uname, lname, policy) \ | |
445 | CLEARPAGEFLAG(uname, lname, policy) | |
f94a62e9 | 446 | |
95ad9755 KS |
447 | #define __PAGEFLAG(uname, lname, policy) \ |
448 | TESTPAGEFLAG(uname, lname, policy) \ | |
449 | __SETPAGEFLAG(uname, lname, policy) \ | |
450 | __CLEARPAGEFLAG(uname, lname, policy) | |
f94a62e9 | 451 | |
95ad9755 KS |
452 | #define TESTSCFLAG(uname, lname, policy) \ |
453 | TESTSETFLAG(uname, lname, policy) \ | |
454 | TESTCLEARFLAG(uname, lname, policy) | |
f94a62e9 | 455 | |
12bbaae7 MWO |
456 | #define FOLIO_TEST_FLAG_FALSE(name) \ |
457 | static inline bool folio_test_##name(const struct folio *folio) \ | |
458 | { return false; } | |
459 | #define FOLIO_SET_FLAG_NOOP(name) \ | |
460 | static inline void folio_set_##name(struct folio *folio) { } | |
461 | #define FOLIO_CLEAR_FLAG_NOOP(name) \ | |
462 | static inline void folio_clear_##name(struct folio *folio) { } | |
463 | #define __FOLIO_SET_FLAG_NOOP(name) \ | |
464 | static inline void __folio_set_##name(struct folio *folio) { } | |
465 | #define __FOLIO_CLEAR_FLAG_NOOP(name) \ | |
466 | static inline void __folio_clear_##name(struct folio *folio) { } | |
467 | #define FOLIO_TEST_SET_FLAG_FALSE(name) \ | |
468 | static inline bool folio_test_set_##name(struct folio *folio) \ | |
469 | { return false; } | |
470 | #define FOLIO_TEST_CLEAR_FLAG_FALSE(name) \ | |
471 | static inline bool folio_test_clear_##name(struct folio *folio) \ | |
472 | { return false; } | |
473 | ||
474 | #define FOLIO_FLAG_FALSE(name) \ | |
475 | FOLIO_TEST_FLAG_FALSE(name) \ | |
476 | FOLIO_SET_FLAG_NOOP(name) \ | |
477 | FOLIO_CLEAR_FLAG_NOOP(name) | |
478 | ||
d389a4a8 | 479 | #define TESTPAGEFLAG_FALSE(uname, lname) \ |
12bbaae7 | 480 | FOLIO_TEST_FLAG_FALSE(lname) \ |
2f3e442c JW |
481 | static inline int Page##uname(const struct page *page) { return 0; } |
482 | ||
d389a4a8 | 483 | #define SETPAGEFLAG_NOOP(uname, lname) \ |
12bbaae7 | 484 | FOLIO_SET_FLAG_NOOP(lname) \ |
8a7a8544 LS |
485 | static inline void SetPage##uname(struct page *page) { } |
486 | ||
d389a4a8 | 487 | #define CLEARPAGEFLAG_NOOP(uname, lname) \ |
12bbaae7 | 488 | FOLIO_CLEAR_FLAG_NOOP(lname) \ |
8a7a8544 LS |
489 | static inline void ClearPage##uname(struct page *page) { } |
490 | ||
d389a4a8 | 491 | #define __CLEARPAGEFLAG_NOOP(uname, lname) \ |
12bbaae7 | 492 | __FOLIO_CLEAR_FLAG_NOOP(lname) \ |
8a7a8544 LS |
493 | static inline void __ClearPage##uname(struct page *page) { } |
494 | ||
d389a4a8 | 495 | #define TESTSETFLAG_FALSE(uname, lname) \ |
12bbaae7 | 496 | FOLIO_TEST_SET_FLAG_FALSE(lname) \ |
2f3e442c JW |
497 | static inline int TestSetPage##uname(struct page *page) { return 0; } |
498 | ||
d389a4a8 | 499 | #define TESTCLEARFLAG_FALSE(uname, lname) \ |
12bbaae7 | 500 | FOLIO_TEST_CLEAR_FLAG_FALSE(lname) \ |
8a7a8544 LS |
501 | static inline int TestClearPage##uname(struct page *page) { return 0; } |
502 | ||
d389a4a8 MWO |
503 | #define PAGEFLAG_FALSE(uname, lname) TESTPAGEFLAG_FALSE(uname, lname) \ |
504 | SETPAGEFLAG_NOOP(uname, lname) CLEARPAGEFLAG_NOOP(uname, lname) | |
2f3e442c | 505 | |
d389a4a8 MWO |
506 | #define TESTSCFLAG_FALSE(uname, lname) \ |
507 | TESTSETFLAG_FALSE(uname, lname) TESTCLEARFLAG_FALSE(uname, lname) | |
2f3e442c | 508 | |
48c935ad | 509 | __PAGEFLAG(Locked, locked, PF_NO_TAIL) |
0d846469 | 510 | FOLIO_FLAG(waiters, FOLIO_HEAD_PAGE) |
093137ea MWO |
511 | FOLIO_FLAG(referenced, FOLIO_HEAD_PAGE) |
512 | FOLIO_TEST_CLEAR_FLAG(referenced, FOLIO_HEAD_PAGE) | |
513 | __FOLIO_SET_FLAG(referenced, FOLIO_HEAD_PAGE) | |
df8c94d1 KS |
514 | PAGEFLAG(Dirty, dirty, PF_HEAD) TESTSCFLAG(Dirty, dirty, PF_HEAD) |
515 | __CLEARPAGEFLAG(Dirty, dirty, PF_HEAD) | |
8cb38fab | 516 | PAGEFLAG(LRU, lru, PF_HEAD) __CLEARPAGEFLAG(LRU, lru, PF_HEAD) |
d25b5bd8 | 517 | TESTCLEARFLAG(LRU, lru, PF_HEAD) |
bf03c806 MWO |
518 | FOLIO_FLAG(active, FOLIO_HEAD_PAGE) |
519 | __FOLIO_CLEAR_FLAG(active, FOLIO_HEAD_PAGE) | |
520 | FOLIO_TEST_CLEAR_FLAG(active, FOLIO_HEAD_PAGE) | |
1899ad18 JW |
521 | PAGEFLAG(Workingset, workingset, PF_HEAD) |
522 | TESTCLEARFLAG(Workingset, workingset, PF_HEAD) | |
df8c94d1 | 523 | PAGEFLAG(Checked, checked, PF_NO_COMPOUND) /* Used by some filesystems */ |
c13985fa KS |
524 | |
525 | /* Xen */ | |
526 | PAGEFLAG(Pinned, pinned, PF_NO_COMPOUND) | |
527 | TESTSCFLAG(Pinned, pinned, PF_NO_COMPOUND) | |
528 | PAGEFLAG(SavePinned, savepinned, PF_NO_COMPOUND); | |
529 | PAGEFLAG(Foreign, foreign, PF_NO_COMPOUND); | |
b877ac98 JG |
530 | PAGEFLAG(XenRemapped, xen_remapped, PF_NO_COMPOUND) |
531 | TESTCLEARFLAG(XenRemapped, xen_remapped, PF_NO_COMPOUND) | |
c13985fa | 532 | |
de09d31d KS |
533 | PAGEFLAG(Reserved, reserved, PF_NO_COMPOUND) |
534 | __CLEARPAGEFLAG(Reserved, reserved, PF_NO_COMPOUND) | |
d483da5b | 535 | __SETPAGEFLAG(Reserved, reserved, PF_NO_COMPOUND) |
0b758280 MWO |
536 | FOLIO_FLAG(swapbacked, FOLIO_HEAD_PAGE) |
537 | __FOLIO_CLEAR_FLAG(swapbacked, FOLIO_HEAD_PAGE) | |
538 | __FOLIO_SET_FLAG(swapbacked, FOLIO_HEAD_PAGE) | |
95ad9755 | 539 | |
266cf658 DH |
540 | /* |
541 | * Private page markings that may be used by the filesystem that owns the page | |
542 | * for its own purposes. | |
704ead2b | 543 | * - PG_private and PG_private_2 cause release_folio() and co to be invoked |
266cf658 | 544 | */ |
2ee08717 | 545 | PAGEFLAG(Private, private, PF_ANY) |
95ad9755 | 546 | PAGEFLAG(Private2, private_2, PF_ANY) TESTSCFLAG(Private2, private_2, PF_ANY) |
266cf658 | 547 | |
02e1960a MWO |
548 | /* owner_2 can be set on tail pages for anon memory */ |
549 | FOLIO_FLAG(owner_2, FOLIO_HEAD_PAGE) | |
550 | ||
6a1e7f77 CL |
551 | /* |
552 | * Only test-and-set exist for PG_writeback. The unconditional operators are | |
553 | * risky: they bypass page accounting. | |
554 | */ | |
225311a4 HY |
555 | TESTPAGEFLAG(Writeback, writeback, PF_NO_TAIL) |
556 | TESTSCFLAG(Writeback, writeback, PF_NO_TAIL) | |
e2f0a0db | 557 | PAGEFLAG(MappedToDisk, mappedtodisk, PF_NO_TAIL) |
6a1e7f77 | 558 | |
579f8290 | 559 | /* PG_readahead is only used for reads; PG_reclaim is only for writes */ |
e2f0a0db KS |
560 | PAGEFLAG(Reclaim, reclaim, PF_NO_TAIL) |
561 | TESTCLEARFLAG(Reclaim, reclaim, PF_NO_TAIL) | |
6f394ee9 MWO |
562 | FOLIO_FLAG(readahead, FOLIO_HEAD_PAGE) |
563 | FOLIO_TEST_CLEAR_FLAG(readahead, FOLIO_HEAD_PAGE) | |
6a1e7f77 CL |
564 | |
565 | #ifdef CONFIG_HIGHMEM | |
1da177e4 | 566 | /* |
6a1e7f77 CL |
567 | * Must use a macro here due to header dependency issues. page_zone() is not |
568 | * available at this point. | |
1da177e4 | 569 | */ |
3ca65c19 | 570 | #define PageHighMem(__p) is_highmem_idx(page_zonenum(__p)) |
00cdf760 | 571 | #define folio_test_highmem(__f) is_highmem_idx(folio_zonenum(__f)) |
6a1e7f77 | 572 | #else |
d389a4a8 | 573 | PAGEFLAG_FALSE(HighMem, highmem) |
6a1e7f77 CL |
574 | #endif |
575 | ||
576 | #ifdef CONFIG_SWAP | |
29cfe755 | 577 | static __always_inline bool folio_test_swapcache(const struct folio *folio) |
6326fec1 | 578 | { |
d389a4a8 | 579 | return folio_test_swapbacked(folio) && |
29cfe755 | 580 | test_bit(PG_swapcache, const_folio_flags(folio, 0)); |
d389a4a8 | 581 | } |
6326fec1 | 582 | |
32f51ead MWO |
583 | FOLIO_SET_FLAG(swapcache, FOLIO_HEAD_PAGE) |
584 | FOLIO_CLEAR_FLAG(swapcache, FOLIO_HEAD_PAGE) | |
6a1e7f77 | 585 | #else |
32f51ead | 586 | FOLIO_FLAG_FALSE(swapcache) |
6a1e7f77 CL |
587 | #endif |
588 | ||
cb29e794 MWO |
589 | FOLIO_FLAG(unevictable, FOLIO_HEAD_PAGE) |
590 | __FOLIO_CLEAR_FLAG(unevictable, FOLIO_HEAD_PAGE) | |
591 | FOLIO_TEST_CLEAR_FLAG(unevictable, FOLIO_HEAD_PAGE) | |
b291f000 | 592 | |
af8e3354 | 593 | #ifdef CONFIG_MMU |
99f86bbd MWO |
594 | FOLIO_FLAG(mlocked, FOLIO_HEAD_PAGE) |
595 | __FOLIO_CLEAR_FLAG(mlocked, FOLIO_HEAD_PAGE) | |
596 | FOLIO_TEST_CLEAR_FLAG(mlocked, FOLIO_HEAD_PAGE) | |
597 | FOLIO_TEST_SET_FLAG(mlocked, FOLIO_HEAD_PAGE) | |
894bc310 | 598 | #else |
99f86bbd MWO |
599 | FOLIO_FLAG_FALSE(mlocked) |
600 | __FOLIO_CLEAR_FLAG_NOOP(mlocked) | |
601 | FOLIO_TEST_CLEAR_FLAG_FALSE(mlocked) | |
602 | FOLIO_TEST_SET_FLAG_FALSE(mlocked) | |
894bc310 LS |
603 | #endif |
604 | ||
d466f2fc | 605 | #ifdef CONFIG_MEMORY_FAILURE |
95ad9755 KS |
606 | PAGEFLAG(HWPoison, hwpoison, PF_ANY) |
607 | TESTSCFLAG(HWPoison, hwpoison, PF_ANY) | |
d466f2fc AK |
608 | #define __PG_HWPOISON (1UL << PG_hwpoison) |
609 | #else | |
d389a4a8 | 610 | PAGEFLAG_FALSE(HWPoison, hwpoison) |
d466f2fc AK |
611 | #define __PG_HWPOISON 0 |
612 | #endif | |
613 | ||
8c9e8381 MWO |
614 | #ifdef CONFIG_PAGE_IDLE_FLAG |
615 | #ifdef CONFIG_64BIT | |
7da8988c MWO |
616 | FOLIO_TEST_FLAG(young, FOLIO_HEAD_PAGE) |
617 | FOLIO_SET_FLAG(young, FOLIO_HEAD_PAGE) | |
618 | FOLIO_TEST_CLEAR_FLAG(young, FOLIO_HEAD_PAGE) | |
619 | FOLIO_FLAG(idle, FOLIO_HEAD_PAGE) | |
33c3fc71 | 620 | #endif |
8c9e8381 MWO |
621 | /* See page_idle.h for !64BIT workaround */ |
622 | #else /* !CONFIG_PAGE_IDLE_FLAG */ | |
623 | FOLIO_FLAG_FALSE(young) | |
624 | FOLIO_TEST_CLEAR_FLAG_FALSE(young) | |
625 | FOLIO_FLAG_FALSE(idle) | |
626 | #endif | |
33c3fc71 | 627 | |
36e66c55 AD |
628 | /* |
629 | * PageReported() is used to track reported free pages within the Buddy | |
630 | * allocator. We can use the non-atomic version of the test and set | |
631 | * operations as both should be shielded with the zone lock to prevent | |
632 | * any possible races on the setting or clearing of the bit. | |
633 | */ | |
634 | __PAGEFLAG(Reported, reported, PF_NO_COMPOUND) | |
635 | ||
66361095 MS |
636 | #ifdef CONFIG_MEMORY_HOTPLUG |
637 | PAGEFLAG(VmemmapSelfHosted, vmemmap_self_hosted, PF_ANY) | |
638 | #else | |
639 | PAGEFLAG_FALSE(VmemmapSelfHosted, vmemmap_self_hosted) | |
640 | #endif | |
641 | ||
e8c6158f | 642 | /* |
06668257 MWO |
643 | * On an anonymous folio mapped into a user virtual memory area, |
644 | * folio->mapping points to its anon_vma, not to a struct address_space; | |
e8c6158f KS |
645 | * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h. |
646 | * | |
647 | * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled, | |
bda807d4 | 648 | * the PAGE_MAPPING_MOVABLE bit may be set along with the PAGE_MAPPING_ANON |
06668257 | 649 | * bit; and then folio->mapping points, not to an anon_vma, but to a private |
e8c6158f KS |
650 | * structure which KSM associates with that merged page. See ksm.h. |
651 | * | |
bda807d4 | 652 | * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is used for non-lru movable |
06668257 | 653 | * page and then folio->mapping points to a struct movable_operations. |
e8c6158f | 654 | * |
06668257 MWO |
655 | * Please note that, confusingly, "folio_mapping" refers to the inode |
656 | * address_space which maps the folio from disk; whereas "folio_mapped" | |
657 | * refers to user virtual address space into which the folio is mapped. | |
44d0fb38 RL |
658 | * |
659 | * For slab pages, since slab reuses the bits in struct page to store its | |
06668257 MWO |
660 | * internal states, the folio->mapping does not exist as such, nor do |
661 | * these flags below. So in order to avoid testing non-existent bits, | |
662 | * please make sure that folio_test_slab(folio) actually evaluates to | |
663 | * false before calling the following functions (e.g., folio_test_anon). | |
664 | * See mm/slab.h. | |
e8c6158f | 665 | */ |
bda807d4 MK |
666 | #define PAGE_MAPPING_ANON 0x1 |
667 | #define PAGE_MAPPING_MOVABLE 0x2 | |
668 | #define PAGE_MAPPING_KSM (PAGE_MAPPING_ANON | PAGE_MAPPING_MOVABLE) | |
669 | #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_MOVABLE) | |
e8c6158f | 670 | |
6061b69b SR |
671 | /* |
672 | * Different with flags above, this flag is used only for fsdax mode. It | |
673 | * indicates that this page->mapping is now under reflink case. | |
674 | */ | |
16900426 | 675 | #define PAGE_MAPPING_DAX_SHARED ((void *)0x1) |
6061b69b | 676 | |
29cfe755 | 677 | static __always_inline bool folio_mapping_flags(const struct folio *folio) |
a9595b30 MWO |
678 | { |
679 | return ((unsigned long)folio->mapping & PAGE_MAPPING_FLAGS) != 0; | |
680 | } | |
681 | ||
64c2e895 | 682 | static __always_inline bool PageMappingFlags(const struct page *page) |
17514574 | 683 | { |
bda807d4 | 684 | return ((unsigned long)page->mapping & PAGE_MAPPING_FLAGS) != 0; |
17514574 MG |
685 | } |
686 | ||
29cfe755 | 687 | static __always_inline bool folio_test_anon(const struct folio *folio) |
e8c6158f | 688 | { |
d389a4a8 MWO |
689 | return ((unsigned long)folio->mapping & PAGE_MAPPING_ANON) != 0; |
690 | } | |
691 | ||
29cfe755 | 692 | static __always_inline bool PageAnon(const struct page *page) |
e8c6158f | 693 | { |
d389a4a8 | 694 | return folio_test_anon(page_folio(page)); |
bda807d4 MK |
695 | } |
696 | ||
b8cecb93 MWO |
697 | static __always_inline bool __folio_test_movable(const struct folio *folio) |
698 | { | |
699 | return ((unsigned long)folio->mapping & PAGE_MAPPING_FLAGS) == | |
700 | PAGE_MAPPING_MOVABLE; | |
701 | } | |
702 | ||
eff201b8 | 703 | static __always_inline bool __PageMovable(const struct page *page) |
bda807d4 MK |
704 | { |
705 | return ((unsigned long)page->mapping & PAGE_MAPPING_FLAGS) == | |
706 | PAGE_MAPPING_MOVABLE; | |
e8c6158f KS |
707 | } |
708 | ||
709 | #ifdef CONFIG_KSM | |
710 | /* | |
711 | * A KSM page is one of those write-protected "shared pages" or "merged pages" | |
712 | * which KSM maps into multiple mms, wherever identical anonymous page content | |
713 | * is found in VM_MERGEABLE vmas. It's a PageAnon page, pointing not to any | |
714 | * anon_vma, but to that page's node of the stable tree. | |
715 | */ | |
29cfe755 | 716 | static __always_inline bool folio_test_ksm(const struct folio *folio) |
e8c6158f | 717 | { |
d389a4a8 | 718 | return ((unsigned long)folio->mapping & PAGE_MAPPING_FLAGS) == |
bda807d4 | 719 | PAGE_MAPPING_KSM; |
e8c6158f | 720 | } |
d389a4a8 | 721 | |
29cfe755 | 722 | static __always_inline bool PageKsm(const struct page *page) |
d389a4a8 MWO |
723 | { |
724 | return folio_test_ksm(page_folio(page)); | |
725 | } | |
e8c6158f | 726 | #else |
d389a4a8 | 727 | TESTPAGEFLAG_FALSE(Ksm, ksm) |
e8c6158f KS |
728 | #endif |
729 | ||
dee3d0be | 730 | u64 stable_page_flags(const struct page *page); |
1a9b5b7f | 731 | |
0410cd84 MWO |
732 | /** |
733 | * folio_xor_flags_has_waiters - Change some folio flags. | |
734 | * @folio: The folio. | |
735 | * @mask: Bits set in this word will be changed. | |
736 | * | |
737 | * This must only be used for flags which are changed with the folio | |
738 | * lock held. For example, it is unsafe to use for PG_dirty as that | |
739 | * can be set without the folio lock held. It can also only be used | |
740 | * on flags which are in the range 0-6 as some of the implementations | |
741 | * only affect those bits. | |
742 | * | |
743 | * Return: Whether there are tasks waiting on the folio. | |
744 | */ | |
745 | static inline bool folio_xor_flags_has_waiters(struct folio *folio, | |
746 | unsigned long mask) | |
747 | { | |
748 | return xor_unlock_is_negative_byte(mask, folio_flags(folio, 0)); | |
749 | } | |
750 | ||
ece01414 MWO |
751 | /** |
752 | * folio_test_uptodate - Is this folio up to date? | |
753 | * @folio: The folio. | |
754 | * | |
755 | * The uptodate flag is set on a folio when every byte in the folio is | |
756 | * at least as new as the corresponding bytes on storage. Anonymous | |
757 | * and CoW folios are always uptodate. If the folio is not uptodate, | |
758 | * some of the bytes in it may be; see the is_partially_uptodate() | |
759 | * address_space operation. | |
760 | */ | |
29cfe755 | 761 | static inline bool folio_test_uptodate(const struct folio *folio) |
0ed361de | 762 | { |
29cfe755 | 763 | bool ret = test_bit(PG_uptodate, const_folio_flags(folio, 0)); |
0ed361de | 764 | /* |
d389a4a8 MWO |
765 | * Must ensure that the data we read out of the folio is loaded |
766 | * _after_ we've loaded folio->flags to check the uptodate bit. | |
767 | * We can skip the barrier if the folio is not uptodate, because | |
0ed361de NP |
768 | * we wouldn't be reading anything from it. |
769 | * | |
d389a4a8 | 770 | * See folio_mark_uptodate() for the other side of the story. |
0ed361de NP |
771 | */ |
772 | if (ret) | |
773 | smp_rmb(); | |
774 | ||
775 | return ret; | |
776 | } | |
777 | ||
2d8b272c | 778 | static inline bool PageUptodate(const struct page *page) |
d389a4a8 MWO |
779 | { |
780 | return folio_test_uptodate(page_folio(page)); | |
781 | } | |
782 | ||
783 | static __always_inline void __folio_mark_uptodate(struct folio *folio) | |
0ed361de NP |
784 | { |
785 | smp_wmb(); | |
d389a4a8 | 786 | __set_bit(PG_uptodate, folio_flags(folio, 0)); |
0ed361de NP |
787 | } |
788 | ||
d389a4a8 | 789 | static __always_inline void folio_mark_uptodate(struct folio *folio) |
2dcea57a | 790 | { |
0ed361de NP |
791 | /* |
792 | * Memory barrier must be issued before setting the PG_uptodate bit, | |
d389a4a8 MWO |
793 | * so that all previous stores issued in order to bring the folio |
794 | * uptodate are actually visible before folio_test_uptodate becomes true. | |
0ed361de NP |
795 | */ |
796 | smp_wmb(); | |
d389a4a8 MWO |
797 | set_bit(PG_uptodate, folio_flags(folio, 0)); |
798 | } | |
799 | ||
800 | static __always_inline void __SetPageUptodate(struct page *page) | |
801 | { | |
802 | __folio_mark_uptodate((struct folio *)page); | |
803 | } | |
804 | ||
805 | static __always_inline void SetPageUptodate(struct page *page) | |
806 | { | |
807 | folio_mark_uptodate((struct folio *)page); | |
0ed361de NP |
808 | } |
809 | ||
d2998c4d | 810 | CLEARPAGEFLAG(Uptodate, uptodate, PF_NO_TAIL) |
1da177e4 | 811 | |
b5612c36 MWO |
812 | void __folio_start_writeback(struct folio *folio, bool keep_write); |
813 | void set_page_writeback(struct page *page); | |
1c8349a1 | 814 | |
f143f1ea MWO |
815 | #define folio_start_writeback(folio) \ |
816 | __folio_start_writeback(folio, false) | |
817 | #define folio_start_writeback_keepwrite(folio) \ | |
818 | __folio_start_writeback(folio, true) | |
1da177e4 | 819 | |
29cfe755 | 820 | static __always_inline bool folio_test_head(const struct folio *folio) |
e7d32485 | 821 | { |
29cfe755 | 822 | return test_bit(PG_head, const_folio_flags(folio, FOLIO_PF_ANY)); |
e7d32485 MS |
823 | } |
824 | ||
ce3467af | 825 | static __always_inline int PageHead(const struct page *page) |
e7d32485 MS |
826 | { |
827 | PF_POISONED_CHECK(page); | |
828 | return test_bit(PG_head, &page->flags) && !page_is_fake_head(page); | |
829 | } | |
830 | ||
831 | __SETPAGEFLAG(Head, head, PF_ANY) | |
832 | __CLEARPAGEFLAG(Head, head, PF_ANY) | |
833 | CLEARPAGEFLAG(Head, head, PF_ANY) | |
e20b8cca | 834 | |
9c325215 MWO |
835 | /** |
836 | * folio_test_large() - Does this folio contain more than one page? | |
837 | * @folio: The folio to test. | |
838 | * | |
839 | * Return: True if the folio is larger than one page. | |
840 | */ | |
29cfe755 | 841 | static inline bool folio_test_large(const struct folio *folio) |
d389a4a8 MWO |
842 | { |
843 | return folio_test_head(folio); | |
844 | } | |
845 | ||
4b0f3261 | 846 | static __always_inline void set_compound_head(struct page *page, struct page *head) |
ad4b3fb7 | 847 | { |
1d798ca3 | 848 | WRITE_ONCE(page->compound_head, (unsigned long)head + 1); |
ad4b3fb7 CD |
849 | } |
850 | ||
4b0f3261 | 851 | static __always_inline void clear_compound_head(struct page *page) |
6a1e7f77 | 852 | { |
1d798ca3 | 853 | WRITE_ONCE(page->compound_head, 0); |
6a1e7f77 | 854 | } |
6d777953 | 855 | |
4e6af67e AA |
856 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
857 | static inline void ClearPageCompound(struct page *page) | |
858 | { | |
1d798ca3 KS |
859 | BUG_ON(!PageHead(page)); |
860 | ClearPageHead(page); | |
4e6af67e | 861 | } |
85edc15a | 862 | FOLIO_FLAG(large_rmappable, FOLIO_SECOND_PAGE) |
8422acdc UA |
863 | FOLIO_TEST_FLAG(partially_mapped, FOLIO_SECOND_PAGE) |
864 | /* | |
865 | * PG_partially_mapped is protected by deferred_split split_queue_lock, | |
866 | * so its safe to use non-atomic set/clear. | |
867 | */ | |
868 | __FOLIO_SET_FLAG(partially_mapped, FOLIO_SECOND_PAGE) | |
869 | __FOLIO_CLEAR_FLAG(partially_mapped, FOLIO_SECOND_PAGE) | |
de53c05f | 870 | #else |
85edc15a | 871 | FOLIO_FLAG_FALSE(large_rmappable) |
8422acdc UA |
872 | FOLIO_TEST_FLAG_FALSE(partially_mapped) |
873 | __FOLIO_SET_FLAG_NOOP(partially_mapped) | |
874 | __FOLIO_CLEAR_FLAG_NOOP(partially_mapped) | |
4e6af67e AA |
875 | #endif |
876 | ||
d2a1a1f0 | 877 | #define PG_head_mask ((1UL << PG_head)) |
dfa7e20c | 878 | |
936a5fe6 | 879 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
71e3aac0 AA |
880 | /* |
881 | * PageHuge() only returns true for hugetlbfs pages, but not for | |
882 | * normal or transparent huge pages. | |
883 | * | |
884 | * PageTransHuge() returns true for both transparent huge and | |
885 | * hugetlbfs pages, but not normal pages. PageTransHuge() can only be | |
886 | * called only in the core VM paths where hugetlbfs pages can't exist. | |
887 | */ | |
29cfe755 | 888 | static inline int PageTransHuge(const struct page *page) |
71e3aac0 | 889 | { |
309381fe | 890 | VM_BUG_ON_PAGE(PageTail(page), page); |
71e3aac0 AA |
891 | return PageHead(page); |
892 | } | |
893 | ||
385de357 DN |
894 | /* |
895 | * PageTransCompound returns true for both transparent huge pages | |
896 | * and hugetlbfs pages, so it should only be called when it's known | |
897 | * that hugetlbfs pages aren't involved. | |
898 | */ | |
29cfe755 | 899 | static inline int PageTransCompound(const struct page *page) |
936a5fe6 AA |
900 | { |
901 | return PageCompound(page); | |
902 | } | |
71e3aac0 | 903 | |
385de357 DN |
904 | /* |
905 | * PageTransTail returns true for both transparent huge pages | |
906 | * and hugetlbfs pages, so it should only be called when it's known | |
907 | * that hugetlbfs pages aren't involved. | |
908 | */ | |
29cfe755 | 909 | static inline int PageTransTail(const struct page *page) |
385de357 DN |
910 | { |
911 | return PageTail(page); | |
912 | } | |
936a5fe6 | 913 | #else |
d389a4a8 MWO |
914 | TESTPAGEFLAG_FALSE(TransHuge, transhuge) |
915 | TESTPAGEFLAG_FALSE(TransCompound, transcompound) | |
916 | TESTPAGEFLAG_FALSE(TransCompoundMap, transcompoundmap) | |
917 | TESTPAGEFLAG_FALSE(TransTail, transtail) | |
936a5fe6 AA |
918 | #endif |
919 | ||
eac96c3e YS |
920 | #if defined(CONFIG_MEMORY_FAILURE) && defined(CONFIG_TRANSPARENT_HUGEPAGE) |
921 | /* | |
922 | * PageHasHWPoisoned indicates that at least one subpage is hwpoisoned in the | |
923 | * compound page. | |
924 | * | |
925 | * This flag is set by hwpoison handler. Cleared by THP split or free page. | |
926 | */ | |
927 | PAGEFLAG(HasHWPoisoned, has_hwpoisoned, PF_SECOND) | |
928 | TESTSCFLAG(HasHWPoisoned, has_hwpoisoned, PF_SECOND) | |
929 | #else | |
e6643593 LT |
930 | PAGEFLAG_FALSE(HasHWPoisoned, has_hwpoisoned) |
931 | TESTSCFLAG_FALSE(HasHWPoisoned, has_hwpoisoned) | |
eac96c3e YS |
932 | #endif |
933 | ||
e8c6158f | 934 | /* |
4ffca5a9 MWO |
935 | * For pages that do not use mapcount, page_type may be used. |
936 | * The low 24 bits of pagetype may be used for your own purposes, as long | |
937 | * as you are careful to not affect the top 8 bits. The low bits of | |
938 | * pagetype will be overwritten when you clear the page_type from the page. | |
e8c6158f | 939 | */ |
ff202303 | 940 | enum pagetype { |
4ffca5a9 MWO |
941 | /* 0x00-0x7f are positive numbers, ie mapcount */ |
942 | /* Reserve 0x80-0xef for mapcount overflow. */ | |
943 | PGTY_buddy = 0xf0, | |
944 | PGTY_offline = 0xf1, | |
945 | PGTY_table = 0xf2, | |
946 | PGTY_guard = 0xf3, | |
947 | PGTY_hugetlb = 0xf4, | |
948 | PGTY_slab = 0xf5, | |
949 | PGTY_zsmalloc = 0xf6, | |
950 | PGTY_unaccepted = 0xf7, | |
951 | ||
952 | PGTY_mapcount_underflow = 0xff | |
ff202303 | 953 | }; |
6e292b9b | 954 | |
e880034c | 955 | static inline bool page_type_has_type(int page_type) |
4c85c0be | 956 | { |
4ffca5a9 | 957 | return page_type < (PGTY_mapcount_underflow << 24); |
4c85c0be HY |
958 | } |
959 | ||
e880034c MWO |
960 | /* This takes a mapcount which is one more than page->_mapcount */ |
961 | static inline bool page_mapcount_is_type(unsigned int mapcount) | |
962 | { | |
963 | return page_type_has_type(mapcount - 1); | |
964 | } | |
965 | ||
966 | static inline bool page_has_type(const struct page *page) | |
144552ff | 967 | { |
4ffca5a9 | 968 | return page_mapcount_is_type(data_race(page->page_type)); |
144552ff AY |
969 | } |
970 | ||
12bbaae7 | 971 | #define FOLIO_TYPE_OPS(lname, fname) \ |
4ffca5a9 | 972 | static __always_inline bool folio_test_##fname(const struct folio *folio) \ |
12bbaae7 | 973 | { \ |
4ffca5a9 | 974 | return data_race(folio->page.page_type >> 24) == PGTY_##lname; \ |
12bbaae7 MWO |
975 | } \ |
976 | static __always_inline void __folio_set_##fname(struct folio *folio) \ | |
977 | { \ | |
4ffca5a9 MWO |
978 | VM_BUG_ON_FOLIO(data_race(folio->page.page_type) != UINT_MAX, \ |
979 | folio); \ | |
980 | folio->page.page_type = (unsigned int)PGTY_##lname << 24; \ | |
12bbaae7 MWO |
981 | } \ |
982 | static __always_inline void __folio_clear_##fname(struct folio *folio) \ | |
983 | { \ | |
984 | VM_BUG_ON_FOLIO(!folio_test_##fname(folio), folio); \ | |
4ffca5a9 | 985 | folio->page.page_type = UINT_MAX; \ |
12bbaae7 MWO |
986 | } |
987 | ||
f7bda0d8 | 988 | #define PAGE_TYPE_OPS(uname, lname, fname) \ |
12bbaae7 | 989 | FOLIO_TYPE_OPS(lname, fname) \ |
f7bda0d8 | 990 | static __always_inline int Page##uname(const struct page *page) \ |
632c0a1a | 991 | { \ |
4ffca5a9 | 992 | return data_race(page->page_type >> 24) == PGTY_##lname; \ |
632c0a1a VD |
993 | } \ |
994 | static __always_inline void __SetPage##uname(struct page *page) \ | |
995 | { \ | |
4ffca5a9 MWO |
996 | VM_BUG_ON_PAGE(data_race(page->page_type) != UINT_MAX, page); \ |
997 | page->page_type = (unsigned int)PGTY_##lname << 24; \ | |
632c0a1a VD |
998 | } \ |
999 | static __always_inline void __ClearPage##uname(struct page *page) \ | |
1000 | { \ | |
1001 | VM_BUG_ON_PAGE(!Page##uname(page), page); \ | |
4ffca5a9 | 1002 | page->page_type = UINT_MAX; \ |
12bbaae7 | 1003 | } |
e8c6158f | 1004 | |
632c0a1a | 1005 | /* |
6e292b9b | 1006 | * PageBuddy() indicates that the page is free and in the buddy system |
632c0a1a VD |
1007 | * (see mm/page_alloc.c). |
1008 | */ | |
f7bda0d8 | 1009 | PAGE_TYPE_OPS(Buddy, buddy, buddy) |
e8c6158f | 1010 | |
632c0a1a | 1011 | /* |
ca215086 DH |
1012 | * PageOffline() indicates that the page is logically offline although the |
1013 | * containing section is online. (e.g. inflated in a balloon driver or | |
1014 | * not onlined when onlining the section). | |
1015 | * The content of these pages is effectively stale. Such pages should not | |
1016 | * be touched (read/write/dump/save) except by their owner. | |
aa218795 | 1017 | * |
503b158f DH |
1018 | * When a memory block gets onlined, all pages are initialized with a |
1019 | * refcount of 1 and PageOffline(). generic_online_page() will | |
1020 | * take care of clearing PageOffline(). | |
1021 | * | |
aa218795 DH |
1022 | * If a driver wants to allow to offline unmovable PageOffline() pages without |
1023 | * putting them back to the buddy, it can do so via the memory notifier by | |
1024 | * decrementing the reference count in MEM_GOING_OFFLINE and incrementing the | |
1025 | * reference count in MEM_CANCEL_OFFLINE. When offlining, the PageOffline() | |
50625744 DH |
1026 | * pages (now with a reference count of zero) are treated like free (unmanaged) |
1027 | * pages, allowing the containing memory block to get offlined. A driver that | |
aa218795 | 1028 | * relies on this feature is aware that re-onlining the memory block will |
503b158f | 1029 | * require not giving them to the buddy via generic_online_page(). |
82840451 | 1030 | * |
50625744 DH |
1031 | * Memory offlining code will not adjust the managed page count for any |
1032 | * PageOffline() pages, treating them like they were never exposed to the | |
1033 | * buddy using generic_online_page(). | |
1034 | * | |
82840451 DH |
1035 | * There are drivers that mark a page PageOffline() and expect there won't be |
1036 | * any further access to page content. PFN walkers that read content of random | |
1037 | * pages should check PageOffline() and synchronize with such drivers using | |
1038 | * page_offline_freeze()/page_offline_thaw(). | |
632c0a1a | 1039 | */ |
f7bda0d8 | 1040 | PAGE_TYPE_OPS(Offline, offline, offline) |
e8c6158f | 1041 | |
82840451 DH |
1042 | extern void page_offline_freeze(void); |
1043 | extern void page_offline_thaw(void); | |
1044 | extern void page_offline_begin(void); | |
1045 | extern void page_offline_end(void); | |
1046 | ||
1d40a5ea MW |
1047 | /* |
1048 | * Marks pages in use as page tables. | |
1049 | */ | |
f7bda0d8 | 1050 | PAGE_TYPE_OPS(Table, table, pgtable) |
1d40a5ea | 1051 | |
3972f6bb VB |
1052 | /* |
1053 | * Marks guardpages used with debug_pagealloc. | |
1054 | */ | |
f7bda0d8 | 1055 | PAGE_TYPE_OPS(Guard, guard, guard) |
3972f6bb | 1056 | |
46df8e73 MWO |
1057 | FOLIO_TYPE_OPS(slab, slab) |
1058 | ||
1059 | /** | |
1060 | * PageSlab - Determine if the page belongs to the slab allocator | |
1061 | * @page: The page to test. | |
1062 | * | |
1063 | * Context: Any context. | |
1064 | * Return: True for slab pages, false for any other kind of page. | |
1065 | */ | |
1066 | static inline bool PageSlab(const struct page *page) | |
1067 | { | |
1068 | return folio_test_slab(page_folio(page)); | |
1069 | } | |
1070 | ||
d99e3140 MWO |
1071 | #ifdef CONFIG_HUGETLB_PAGE |
1072 | FOLIO_TYPE_OPS(hugetlb, hugetlb) | |
1073 | #else | |
1074 | FOLIO_TEST_FLAG_FALSE(hugetlb) | |
1075 | #endif | |
1076 | ||
43d746dc DH |
1077 | PAGE_TYPE_OPS(Zsmalloc, zsmalloc, zsmalloc) |
1078 | ||
310183de KS |
1079 | /* |
1080 | * Mark pages that has to be accepted before touched for the first time. | |
1081 | * | |
1082 | * Serialized with zone lock. | |
1083 | */ | |
1084 | PAGE_TYPE_OPS(Unaccepted, unaccepted, unaccepted) | |
1085 | ||
d99e3140 MWO |
1086 | /** |
1087 | * PageHuge - Determine if the page belongs to hugetlbfs | |
1088 | * @page: The page to test. | |
1089 | * | |
1090 | * Context: Any context. | |
1091 | * Return: True for hugetlbfs pages, false for anon pages or pages | |
1092 | * belonging to other filesystems. | |
1093 | */ | |
1094 | static inline bool PageHuge(const struct page *page) | |
1095 | { | |
1096 | return folio_test_hugetlb(page_folio(page)); | |
1097 | } | |
1098 | ||
1099 | /* | |
1100 | * Check if a page is currently marked HWPoisoned. Note that this check is | |
1101 | * best effort only and inherently racy: there is no way to synchronize with | |
1102 | * failing hardware. | |
1103 | */ | |
8682a7be | 1104 | static inline bool is_page_hwpoison(const struct page *page) |
d99e3140 | 1105 | { |
8682a7be MWO |
1106 | const struct folio *folio; |
1107 | ||
d99e3140 MWO |
1108 | if (PageHWPoison(page)) |
1109 | return true; | |
8682a7be MWO |
1110 | folio = page_folio(page); |
1111 | return folio_test_hugetlb(folio) && PageHWPoison(&folio->page); | |
d99e3140 MWO |
1112 | } |
1113 | ||
2ace5a67 | 1114 | bool is_free_buddy_page(const struct page *page); |
832fc1de | 1115 | |
356ea386 | 1116 | PAGEFLAG(Isolated, isolated, PF_ANY); |
bda807d4 | 1117 | |
29cfe755 | 1118 | static __always_inline int PageAnonExclusive(const struct page *page) |
78fbe906 DH |
1119 | { |
1120 | VM_BUG_ON_PGFLAGS(!PageAnon(page), page); | |
c0bff412 PX |
1121 | /* |
1122 | * HugeTLB stores this information on the head page; THP keeps it per | |
1123 | * page | |
1124 | */ | |
1125 | if (PageHuge(page)) | |
1126 | page = compound_head(page); | |
78fbe906 DH |
1127 | return test_bit(PG_anon_exclusive, &PF_ANY(page, 1)->flags); |
1128 | } | |
1129 | ||
1130 | static __always_inline void SetPageAnonExclusive(struct page *page) | |
1131 | { | |
1132 | VM_BUG_ON_PGFLAGS(!PageAnon(page) || PageKsm(page), page); | |
1133 | VM_BUG_ON_PGFLAGS(PageHuge(page) && !PageHead(page), page); | |
1134 | set_bit(PG_anon_exclusive, &PF_ANY(page, 1)->flags); | |
1135 | } | |
1136 | ||
1137 | static __always_inline void ClearPageAnonExclusive(struct page *page) | |
1138 | { | |
1139 | VM_BUG_ON_PGFLAGS(!PageAnon(page) || PageKsm(page), page); | |
1140 | VM_BUG_ON_PGFLAGS(PageHuge(page) && !PageHead(page), page); | |
1141 | clear_bit(PG_anon_exclusive, &PF_ANY(page, 1)->flags); | |
1142 | } | |
1143 | ||
1144 | static __always_inline void __ClearPageAnonExclusive(struct page *page) | |
1145 | { | |
1146 | VM_BUG_ON_PGFLAGS(!PageAnon(page), page); | |
1147 | VM_BUG_ON_PGFLAGS(PageHuge(page) && !PageHead(page), page); | |
1148 | __clear_bit(PG_anon_exclusive, &PF_ANY(page, 1)->flags); | |
1149 | } | |
1150 | ||
af8e3354 | 1151 | #ifdef CONFIG_MMU |
d2a1a1f0 | 1152 | #define __PG_MLOCKED (1UL << PG_mlocked) |
33925b25 | 1153 | #else |
b291f000 | 1154 | #define __PG_MLOCKED 0 |
894bc310 LS |
1155 | #endif |
1156 | ||
dfa7e20c RA |
1157 | /* |
1158 | * Flags checked when a page is freed. Pages being freed should not have | |
4be408ce | 1159 | * these flags set. If they are, there is a problem. |
dfa7e20c | 1160 | */ |
6326fec1 NP |
1161 | #define PAGE_FLAGS_CHECK_AT_FREE \ |
1162 | (1UL << PG_lru | 1UL << PG_locked | \ | |
1163 | 1UL << PG_private | 1UL << PG_private_2 | \ | |
1164 | 1UL << PG_writeback | 1UL << PG_reserved | \ | |
46df8e73 | 1165 | 1UL << PG_active | \ |
ec1c86b2 | 1166 | 1UL << PG_unevictable | __PG_MLOCKED | LRU_GEN_MASK) |
dfa7e20c RA |
1167 | |
1168 | /* | |
1169 | * Flags checked when a page is prepped for return by the page allocator. | |
4be408ce | 1170 | * Pages being prepped should not have these flags set. If they are set, |
79f4b7bf | 1171 | * there has been a kernel bug or struct page corruption. |
f4c18e6f NH |
1172 | * |
1173 | * __PG_HWPOISON is exceptional because it needs to be kept beyond page's | |
1174 | * alloc-free cycle to prevent from reusing the page. | |
dfa7e20c | 1175 | */ |
f4c18e6f | 1176 | #define PAGE_FLAGS_CHECK_AT_PREP \ |
ec1c86b2 | 1177 | ((PAGEFLAGS_MASK & ~__PG_HWPOISON) | LRU_GEN_MASK | LRU_REFS_MASK) |
dfa7e20c | 1178 | |
9c5ccf2d MWO |
1179 | /* |
1180 | * Flags stored in the second page of a compound page. They may overlap | |
1181 | * the CHECK_AT_FREE flags above, so need to be cleared. | |
1182 | */ | |
1183 | #define PAGE_FLAGS_SECOND \ | |
ebc1baf5 | 1184 | (0xffUL /* order */ | 1UL << PG_has_hwpoisoned | \ |
8422acdc | 1185 | 1UL << PG_large_rmappable | 1UL << PG_partially_mapped) |
9c5ccf2d | 1186 | |
edcf4748 | 1187 | #define PAGE_FLAGS_PRIVATE \ |
d2a1a1f0 | 1188 | (1UL << PG_private | 1UL << PG_private_2) |
266cf658 | 1189 | /** |
6dc15138 MWO |
1190 | * folio_has_private - Determine if folio has private stuff |
1191 | * @folio: The folio to be checked | |
266cf658 | 1192 | * |
6dc15138 | 1193 | * Determine if a folio has private stuff, indicating that release routines |
266cf658 DH |
1194 | * should be invoked upon it. |
1195 | */ | |
6dc15138 | 1196 | static inline int folio_has_private(const struct folio *folio) |
edcf4748 | 1197 | { |
6dc15138 | 1198 | return !!(folio->flags & PAGE_FLAGS_PRIVATE); |
d389a4a8 MWO |
1199 | } |
1200 | ||
95ad9755 KS |
1201 | #undef PF_ANY |
1202 | #undef PF_HEAD | |
1203 | #undef PF_NO_TAIL | |
1204 | #undef PF_NO_COMPOUND | |
a08d93e5 | 1205 | #undef PF_SECOND |
edcf4748 | 1206 | #endif /* !__GENERATING_BOUNDS_H */ |
266cf658 | 1207 | |
1da177e4 | 1208 | #endif /* PAGE_FLAGS_H */ |