Blame view
mm/readahead.c
17.3 KB
457c89965 treewide: Add SPD... |
1 |
// SPDX-License-Identifier: GPL-2.0-only |
1da177e4c Linux-2.6.12-rc2 |
2 3 4 5 6 |
/* * mm/readahead.c - address_space-level file readahead. * * Copyright (C) 2002, Linus Torvalds * |
e1f8e8744 Remove Andrew Mor... |
7 |
* 09Apr2002 Andrew Morton |
1da177e4c Linux-2.6.12-rc2 |
8 9 10 11 |
* Initial version. */ #include <linux/kernel.h> |
11bd969fd mm: silently skip... |
12 |
#include <linux/dax.h> |
5a0e3ad6a include cleanup: ... |
13 |
#include <linux/gfp.h> |
b95f1b31b mm: Map most file... |
14 |
#include <linux/export.h> |
1da177e4c Linux-2.6.12-rc2 |
15 16 |
#include <linux/blkdev.h> #include <linux/backing-dev.h> |
8bde37f08 [PATCH] io-accoun... |
17 |
#include <linux/task_io_accounting_ops.h> |
1da177e4c Linux-2.6.12-rc2 |
18 |
#include <linux/pagevec.h> |
f5ff8422b Fix warnings with... |
19 |
#include <linux/pagemap.h> |
782182e53 mm: move readahea... |
20 21 |
#include <linux/syscalls.h> #include <linux/file.h> |
d72ee9111 mm: move lru_to_p... |
22 |
#include <linux/mm_inline.h> |
ca47e8c72 mm: skip readahea... |
23 |
#include <linux/blk-cgroup.h> |
3d8f76153 vfs: implement re... |
24 |
#include <linux/fadvise.h> |
f2c817bed mm: use memalloc_... |
25 |
#include <linux/sched/mm.h> |
1da177e4c Linux-2.6.12-rc2 |
26 |
|
29f175d12 mm/readahead.c: i... |
27 |
#include "internal.h" |
1da177e4c Linux-2.6.12-rc2 |
28 29 30 31 32 33 34 |
/* * Initialise a struct file's readahead state. Assumes that the caller has * memset *ra to zero. */ void file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping) { |
de1414a65 fs: export inode_... |
35 |
ra->ra_pages = inode_to_bdi(mapping->host)->ra_pages; |
f4e6b498d readahead: combin... |
36 |
ra->prev_pos = -1; |
1da177e4c Linux-2.6.12-rc2 |
37 |
} |
d41cc702c [GFS2] Export fil... |
38 |
EXPORT_SYMBOL_GPL(file_ra_state_init); |
1da177e4c Linux-2.6.12-rc2 |
39 |
|
03fb3d2af FS-Cache: Release... |
40 41 |
/* * see if a page needs releasing upon read_cache_pages() failure |
266cf658e FS-Cache: Recruit... |
42 43 44 45 |
* - the caller of read_cache_pages() may have set PG_private or PG_fscache * before calling, such as the NFS fs marking pages that are cached locally * on disk, thus we need to give the fs a chance to clean up in the event of * an error |
03fb3d2af FS-Cache: Release... |
46 47 48 49 |
*/ static void read_cache_pages_invalidate_page(struct address_space *mapping, struct page *page) { |
266cf658e FS-Cache: Recruit... |
50 |
if (page_has_private(page)) { |
03fb3d2af FS-Cache: Release... |
51 52 53 |
if (!trylock_page(page)) BUG(); page->mapping = mapping; |
09cbfeaf1 mm, fs: get rid o... |
54 |
do_invalidatepage(page, 0, PAGE_SIZE); |
03fb3d2af FS-Cache: Release... |
55 56 57 |
page->mapping = NULL; unlock_page(page); } |
09cbfeaf1 mm, fs: get rid o... |
58 |
put_page(page); |
03fb3d2af FS-Cache: Release... |
59 60 61 62 63 64 65 66 67 68 69 |
} /* * release a list of pages, invalidating them first if need be */ static void read_cache_pages_invalidate_pages(struct address_space *mapping, struct list_head *pages) { struct page *victim; while (!list_empty(pages)) { |
c8ad6302c mm/readahead.c, m... |
70 |
victim = lru_to_page(pages); |
03fb3d2af FS-Cache: Release... |
71 72 73 74 |
list_del(&victim->lru); read_cache_pages_invalidate_page(mapping, victim); } } |
1da177e4c Linux-2.6.12-rc2 |
75 |
/** |
bd40cddae [PATCH] kernel-do... |
76 |
* read_cache_pages - populate an address space with some pages & start reads against them |
1da177e4c Linux-2.6.12-rc2 |
77 78 79 80 81 82 83 |
* @mapping: the address_space * @pages: The address of a list_head which contains the target pages. These * pages have their ->index populated and are otherwise uninitialised. * @filler: callback routine for filling a single page. * @data: private data for the callback routine. * * Hides the details of the LRU cache etc from the filesystems. |
a862f68a8 docs/core-api/mm:... |
84 85 |
* * Returns: %0 on success, error return by @filler otherwise |
1da177e4c Linux-2.6.12-rc2 |
86 87 88 89 90 |
*/ int read_cache_pages(struct address_space *mapping, struct list_head *pages, int (*filler)(void *, struct page *), void *data) { struct page *page; |
1da177e4c Linux-2.6.12-rc2 |
91 |
int ret = 0; |
1da177e4c Linux-2.6.12-rc2 |
92 |
while (!list_empty(pages)) { |
c8ad6302c mm/readahead.c, m... |
93 |
page = lru_to_page(pages); |
1da177e4c Linux-2.6.12-rc2 |
94 |
list_del(&page->lru); |
063d99b4f mm, fs: obey gfp_... |
95 |
if (add_to_page_cache_lru(page, mapping, page->index, |
8a5c743e3 mm, memcg: use co... |
96 |
readahead_gfp_mask(mapping))) { |
03fb3d2af FS-Cache: Release... |
97 |
read_cache_pages_invalidate_page(mapping, page); |
1da177e4c Linux-2.6.12-rc2 |
98 99 |
continue; } |
09cbfeaf1 mm, fs: get rid o... |
100 |
put_page(page); |
eb2be1893 mm: buffered writ... |
101 |
|
1da177e4c Linux-2.6.12-rc2 |
102 |
ret = filler(data, page); |
eb2be1893 mm: buffered writ... |
103 |
if (unlikely(ret)) { |
03fb3d2af FS-Cache: Release... |
104 |
read_cache_pages_invalidate_pages(mapping, pages); |
1da177e4c Linux-2.6.12-rc2 |
105 106 |
break; } |
09cbfeaf1 mm, fs: get rid o... |
107 |
task_io_account_read(PAGE_SIZE); |
1da177e4c Linux-2.6.12-rc2 |
108 |
} |
1da177e4c Linux-2.6.12-rc2 |
109 110 111 112 |
return ret; } EXPORT_SYMBOL(read_cache_pages); |
a4d965366 mm: use readahead... |
113 |
static void read_pages(struct readahead_control *rac, struct list_head *pages, |
c1f6925e1 mm: put readahead... |
114 |
bool skip_page) |
1da177e4c Linux-2.6.12-rc2 |
115 |
{ |
a4d965366 mm: use readahead... |
116 |
const struct address_space_operations *aops = rac->mapping->a_ops; |
c1f6925e1 mm: put readahead... |
117 |
struct page *page; |
5b417b187 read-ahead: use p... |
118 |
struct blk_plug plug; |
1da177e4c Linux-2.6.12-rc2 |
119 |
|
a4d965366 mm: use readahead... |
120 |
if (!readahead_count(rac)) |
c1f6925e1 mm: put readahead... |
121 |
goto out; |
ad4ae1c73 mm: move readahea... |
122 |
|
5b417b187 read-ahead: use p... |
123 |
blk_start_plug(&plug); |
8151b4c8b mm: add readahead... |
124 125 126 127 128 129 130 131 |
if (aops->readahead) { aops->readahead(rac); /* Clean up the remaining pages */ while ((page = readahead_page(rac))) { unlock_page(page); put_page(page); } } else if (aops->readpages) { |
a4d965366 mm: use readahead... |
132 133 |
aops->readpages(rac->file, rac->mapping, pages, readahead_count(rac)); |
029e332ea [PATCH] Cleanup r... |
134 135 |
/* Clean up the remaining pages */ put_pages_list(pages); |
c1f6925e1 mm: put readahead... |
136 137 138 139 |
rac->_index += rac->_nr_pages; rac->_nr_pages = 0; } else { while ((page = readahead_page(rac))) { |
a4d965366 mm: use readahead... |
140 |
aops->readpage(rac->file, page); |
c1f6925e1 mm: put readahead... |
141 142 |
put_page(page); } |
1da177e4c Linux-2.6.12-rc2 |
143 |
} |
5b417b187 read-ahead: use p... |
144 |
|
5b417b187 read-ahead: use p... |
145 |
blk_finish_plug(&plug); |
ad4ae1c73 mm: move readahea... |
146 147 |
BUG_ON(!list_empty(pages)); |
c1f6925e1 mm: put readahead... |
148 149 150 151 152 |
BUG_ON(readahead_count(rac)); out: if (skip_page) rac->_index++; |
1da177e4c Linux-2.6.12-rc2 |
153 |
} |
2c684234d mm: add page_cach... |
154 |
/** |
73bb49da5 mm/readahead: mak... |
155 156 |
* page_cache_ra_unbounded - Start unchecked readahead. * @ractl: Readahead control. |
2c684234d mm: add page_cach... |
157 158 159 160 161 162 163 164 165 166 |
* @nr_to_read: The number of pages to read. * @lookahead_size: Where to start the next readahead. * * This function is for filesystems to call when they want to start * readahead beyond a file's stated i_size. This is almost certainly * not the function you want to call. Use page_cache_async_readahead() * or page_cache_sync_readahead() instead. * * Context: File is referenced by caller. Mutexes may be held by caller. * May sleep, but will not reenter filesystem to reclaim memory. |
1da177e4c Linux-2.6.12-rc2 |
167 |
*/ |
73bb49da5 mm/readahead: mak... |
168 169 |
void page_cache_ra_unbounded(struct readahead_control *ractl, unsigned long nr_to_read, unsigned long lookahead_size) |
1da177e4c Linux-2.6.12-rc2 |
170 |
{ |
73bb49da5 mm/readahead: mak... |
171 172 |
struct address_space *mapping = ractl->mapping; unsigned long index = readahead_index(ractl); |
1da177e4c Linux-2.6.12-rc2 |
173 |
LIST_HEAD(page_pool); |
8a5c743e3 mm, memcg: use co... |
174 |
gfp_t gfp_mask = readahead_gfp_mask(mapping); |
c2c7ad74b mm: rename readah... |
175 |
unsigned long i; |
1da177e4c Linux-2.6.12-rc2 |
176 177 |
/* |
f2c817bed mm: use memalloc_... |
178 179 180 181 182 183 184 185 186 187 188 189 |
* Partway through the readahead operation, we will have added * locked pages to the page cache, but will not yet have submitted * them for I/O. Adding another page may need to allocate memory, * which can trigger memory reclaim. Telling the VM we're in * the middle of a filesystem operation will cause it to not * touch file-backed pages, preventing a deadlock. Most (all?) * filesystems already specify __GFP_NOFS in their mapping's * gfp_mask, but let's be explicit here. */ unsigned int nofs = memalloc_nofs_save(); /* |
1da177e4c Linux-2.6.12-rc2 |
190 191 |
* Preallocate as many pages as we will need. */ |
c2c7ad74b mm: rename readah... |
192 |
for (i = 0; i < nr_to_read; i++) { |
b0f31d78c mm: move end_inde... |
193 |
struct page *page = xa_load(&mapping->i_pages, index + i); |
1da177e4c Linux-2.6.12-rc2 |
194 |
|
73bb49da5 mm/readahead: mak... |
195 |
BUG_ON(index + i != ractl->_index + ractl->_nr_pages); |
c1f6925e1 mm: put readahead... |
196 |
|
3159f943a xarray: Replace e... |
197 |
if (page && !xa_is_value(page)) { |
b3751e6ab mm: split ->readp... |
198 |
/* |
2d8163e48 mm: document why ... |
199 200 201 202 203 204 |
* Page already present? Kick off the current batch * of contiguous pages before continuing with the * next batch. This page may be the one we would * have intended to mark as Readahead, but we don't * have a stable reference to this page, and it's * not worth getting one just for that. |
b3751e6ab mm: split ->readp... |
205 |
*/ |
73bb49da5 mm/readahead: mak... |
206 |
read_pages(ractl, &page_pool, true); |
1da177e4c Linux-2.6.12-rc2 |
207 |
continue; |
b3751e6ab mm: split ->readp... |
208 |
} |
1da177e4c Linux-2.6.12-rc2 |
209 |
|
8a5c743e3 mm, memcg: use co... |
210 |
page = __page_cache_alloc(gfp_mask); |
1da177e4c Linux-2.6.12-rc2 |
211 212 |
if (!page) break; |
c1f6925e1 mm: put readahead... |
213 214 215 216 217 218 |
if (mapping->a_ops->readpages) { page->index = index + i; list_add(&page->lru, &page_pool); } else if (add_to_page_cache_lru(page, mapping, index + i, gfp_mask) < 0) { put_page(page); |
73bb49da5 mm/readahead: mak... |
219 |
read_pages(ractl, &page_pool, true); |
c1f6925e1 mm: put readahead... |
220 221 |
continue; } |
c2c7ad74b mm: rename readah... |
222 |
if (i == nr_to_read - lookahead_size) |
46fc3e7b4 readahead: add lo... |
223 |
SetPageReadahead(page); |
73bb49da5 mm/readahead: mak... |
224 |
ractl->_nr_pages++; |
1da177e4c Linux-2.6.12-rc2 |
225 |
} |
1da177e4c Linux-2.6.12-rc2 |
226 227 228 229 230 231 |
/* * Now start the IO. We ignore I/O errors - if the page is not * uptodate then the caller will launch readpage again, and * will then handle the error. */ |
73bb49da5 mm/readahead: mak... |
232 |
read_pages(ractl, &page_pool, false); |
f2c817bed mm: use memalloc_... |
233 |
memalloc_nofs_restore(nofs); |
1da177e4c Linux-2.6.12-rc2 |
234 |
} |
73bb49da5 mm/readahead: mak... |
235 |
EXPORT_SYMBOL_GPL(page_cache_ra_unbounded); |
2c684234d mm: add page_cach... |
236 237 |
/* |
8238287ea mm/readahead: mak... |
238 |
* do_page_cache_ra() actually reads a chunk of disk. It allocates |
2c684234d mm: add page_cach... |
239 240 241 242 |
* the pages first, then submits them for I/O. This avoids the very bad * behaviour which would occur if page allocations are causing VM writeback. * We really don't want to intermingle reads and writes like that. */ |
8238287ea mm/readahead: mak... |
243 244 |
void do_page_cache_ra(struct readahead_control *ractl, unsigned long nr_to_read, unsigned long lookahead_size) |
2c684234d mm: add page_cach... |
245 |
{ |
8238287ea mm/readahead: mak... |
246 247 |
struct inode *inode = ractl->mapping->host; unsigned long index = readahead_index(ractl); |
2c684234d mm: add page_cach... |
248 249 250 251 252 253 254 255 256 257 258 259 |
loff_t isize = i_size_read(inode); pgoff_t end_index; /* The last page we want to read */ if (isize == 0) return; end_index = (isize - 1) >> PAGE_SHIFT; if (index > end_index) return; /* Don't read past the page containing the last byte of the file */ if (nr_to_read > end_index - index) nr_to_read = end_index - index + 1; |
8238287ea mm/readahead: mak... |
260 |
page_cache_ra_unbounded(ractl, nr_to_read, lookahead_size); |
2c684234d mm: add page_cach... |
261 |
} |
1da177e4c Linux-2.6.12-rc2 |
262 263 264 265 266 |
/* * Chunk the readahead into 2 megabyte units, so that we don't pin too much * memory at once. */ |
7b3df3b9a mm/readahead: pas... |
267 |
void force_page_cache_ra(struct readahead_control *ractl, |
b1647dc0d mm/readahead: pas... |
268 |
struct file_ra_state *ra, unsigned long nr_to_read) |
1da177e4c Linux-2.6.12-rc2 |
269 |
{ |
7b3df3b9a mm/readahead: pas... |
270 |
struct address_space *mapping = ractl->mapping; |
9491ae4aa mm: don't cap req... |
271 |
struct backing_dev_info *bdi = inode_to_bdi(mapping->host); |
7b3df3b9a mm/readahead: pas... |
272 |
unsigned long max_pages, index; |
9491ae4aa mm: don't cap req... |
273 |
|
8151b4c8b mm: add readahead... |
274 275 |
if (unlikely(!mapping->a_ops->readpage && !mapping->a_ops->readpages && !mapping->a_ops->readahead)) |
9a42823a1 mm: return void f... |
276 |
return; |
1da177e4c Linux-2.6.12-rc2 |
277 |
|
9491ae4aa mm: don't cap req... |
278 279 280 281 |
/* * If the request exceeds the readahead window, allow the read to * be up to the optimal hardware IO size */ |
7b3df3b9a mm/readahead: pas... |
282 |
index = readahead_index(ractl); |
9491ae4aa mm: don't cap req... |
283 |
max_pages = max_t(unsigned long, bdi->io_pages, ra->ra_pages); |
7b3df3b9a mm/readahead: pas... |
284 |
nr_to_read = min_t(unsigned long, nr_to_read, max_pages); |
1da177e4c Linux-2.6.12-rc2 |
285 |
while (nr_to_read) { |
09cbfeaf1 mm, fs: get rid o... |
286 |
unsigned long this_chunk = (2 * 1024 * 1024) / PAGE_SIZE; |
1da177e4c Linux-2.6.12-rc2 |
287 288 289 |
if (this_chunk > nr_to_read) this_chunk = nr_to_read; |
7b3df3b9a mm/readahead: pas... |
290 291 |
ractl->_index = index; do_page_cache_ra(ractl, this_chunk, 0); |
58d5640eb mm/readahead.c: f... |
292 |
|
08eb9658a mm: rename variou... |
293 |
index += this_chunk; |
1da177e4c Linux-2.6.12-rc2 |
294 295 |
nr_to_read -= this_chunk; } |
1da177e4c Linux-2.6.12-rc2 |
296 |
} |
5ce1110b9 readahead: data s... |
297 |
/* |
c743d96b6 readahead: remove... |
298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 |
* Set the initial window size, round to next power of 2 and square * for small size, x 4 for medium, and x 2 for large * for 128k (32 page) max ra * 1-8 page = 32k initial, > 8 page = 128k initial */ static unsigned long get_init_ra_size(unsigned long size, unsigned long max) { unsigned long newsize = roundup_pow_of_two(size); if (newsize <= max / 32) newsize = newsize * 4; else if (newsize <= max / 4) newsize = newsize * 2; else newsize = max; return newsize; } /* |
122a21d11 readahead: on-dem... |
318 319 320 |
* Get the previous window size, ramp it up, and * return it as the new window size. */ |
c743d96b6 readahead: remove... |
321 |
static unsigned long get_next_ra_size(struct file_ra_state *ra, |
20ff1c950 mm/readahead.c: s... |
322 |
unsigned long max) |
122a21d11 readahead: on-dem... |
323 |
{ |
f9acc8c7b readahead: sanify... |
324 |
unsigned long cur = ra->size; |
122a21d11 readahead: on-dem... |
325 326 |
if (cur < max / 16) |
20ff1c950 mm/readahead.c: s... |
327 328 329 330 |
return 4 * cur; if (cur <= max / 2) return 2 * cur; return max; |
122a21d11 readahead: on-dem... |
331 332 333 334 335 336 337 338 |
} /* * On-demand readahead design. * * The fields in struct file_ra_state represent the most-recently-executed * readahead attempt: * |
f9acc8c7b readahead: sanify... |
339 340 341 342 |
* |<----- async_size ---------| * |------------------- size -------------------->| * |==================#===========================| * ^start ^page marked with PG_readahead |
122a21d11 readahead: on-dem... |
343 344 345 346 |
* * To overlap application thinking time and disk I/O time, we do * `readahead pipelining': Do not wait until the application consumed all * readahead pages and stalled on the missing page at readahead_index; |
f9acc8c7b readahead: sanify... |
347 348 349 |
* Instead, submit an asynchronous readahead I/O as soon as there are * only async_size pages left in the readahead window. Normally async_size * will be equal to size, for maximum pipelining. |
122a21d11 readahead: on-dem... |
350 351 352 |
* * In interleaved sequential reads, concurrent streams on the same fd can * be invalidating each other's readahead state. So we flag the new readahead |
f9acc8c7b readahead: sanify... |
353 |
* page at (start+size-async_size) with PG_readahead, and use it as readahead |
122a21d11 readahead: on-dem... |
354 355 356 |
* indicator. The flag won't be set on already cached pages, to avoid the * readahead-for-nothing fuss, saving pointless page cache lookups. * |
f4e6b498d readahead: combin... |
357 |
* prev_pos tracks the last visited byte in the _previous_ read request. |
122a21d11 readahead: on-dem... |
358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 |
* It should be maintained by the caller, and will be used for detecting * small random reads. Note that the readahead algorithm checks loosely * for sequential patterns. Hence interleaved reads might be served as * sequential ones. * * There is a special-case: if the first page which the application tries to * read happens to be the first page of the file, it is assumed that a linear * read is about to happen and the window is immediately set to the initial size * based on I/O request size and the max_readahead. * * The code ramps up the readahead size aggressively at first, but slow down as * it approaches max_readhead. */ /* |
08eb9658a mm: rename variou... |
373 |
* Count contiguously cached pages from @index-1 to @index-@max, |
10be0b372 readahead: introd... |
374 375 376 377 378 |
* this count is a conservative estimation of * - length of the sequential read sequence, or * - thrashing threshold in memory tight systems */ static pgoff_t count_history_pages(struct address_space *mapping, |
08eb9658a mm: rename variou... |
379 |
pgoff_t index, unsigned long max) |
10be0b372 readahead: introd... |
380 381 382 383 |
{ pgoff_t head; rcu_read_lock(); |
08eb9658a mm: rename variou... |
384 |
head = page_cache_prev_miss(mapping, index - 1, max); |
10be0b372 readahead: introd... |
385 |
rcu_read_unlock(); |
08eb9658a mm: rename variou... |
386 |
return index - 1 - head; |
10be0b372 readahead: introd... |
387 388 389 390 391 392 393 |
} /* * page cache context based read-ahead */ static int try_context_readahead(struct address_space *mapping, struct file_ra_state *ra, |
08eb9658a mm: rename variou... |
394 |
pgoff_t index, |
10be0b372 readahead: introd... |
395 396 397 398 |
unsigned long req_size, unsigned long max) { pgoff_t size; |
08eb9658a mm: rename variou... |
399 |
size = count_history_pages(mapping, index, max); |
10be0b372 readahead: introd... |
400 401 |
/* |
2cad40180 readahead: make c... |
402 |
* not enough history pages: |
10be0b372 readahead: introd... |
403 404 |
* it could be a random read */ |
2cad40180 readahead: make c... |
405 |
if (size <= req_size) |
10be0b372 readahead: introd... |
406 407 408 409 410 411 |
return 0; /* * starts from beginning of file: * it is a strong indication of long-run stream (or whole-file-read) */ |
08eb9658a mm: rename variou... |
412 |
if (size >= index) |
10be0b372 readahead: introd... |
413 |
size *= 2; |
08eb9658a mm: rename variou... |
414 |
ra->start = index; |
2cad40180 readahead: make c... |
415 416 |
ra->size = min(size + req_size, max); ra->async_size = 1; |
10be0b372 readahead: introd... |
417 418 419 420 421 |
return 1; } /* |
122a21d11 readahead: on-dem... |
422 423 |
* A minimal readahead algorithm for trivial sequential/random reads. */ |
6e4af69ae mm/readahead: mak... |
424 425 |
static void ondemand_readahead(struct readahead_control *ractl, struct file_ra_state *ra, bool hit_readahead_marker, |
9a42823a1 mm: return void f... |
426 |
unsigned long req_size) |
122a21d11 readahead: on-dem... |
427 |
{ |
6e4af69ae mm/readahead: mak... |
428 |
struct backing_dev_info *bdi = inode_to_bdi(ractl->mapping->host); |
9491ae4aa mm: don't cap req... |
429 |
unsigned long max_pages = ra->ra_pages; |
dc30b96ab readahead: strict... |
430 |
unsigned long add_pages; |
6e4af69ae mm/readahead: mak... |
431 |
unsigned long index = readahead_index(ractl); |
08eb9658a mm: rename variou... |
432 |
pgoff_t prev_index; |
045a2529a readahead: move t... |
433 434 |
/* |
9491ae4aa mm: don't cap req... |
435 436 437 438 439 440 441 |
* If the request exceeds the readahead window, allow the read to * be up to the optimal hardware IO size */ if (req_size > max_pages && bdi->io_pages > max_pages) max_pages = min(req_size, bdi->io_pages); /* |
045a2529a readahead: move t... |
442 443 |
* start of file */ |
08eb9658a mm: rename variou... |
444 |
if (!index) |
045a2529a readahead: move t... |
445 |
goto initial_readahead; |
122a21d11 readahead: on-dem... |
446 447 |
/* |
08eb9658a mm: rename variou... |
448 |
* It's the expected callback index, assume sequential access. |
122a21d11 readahead: on-dem... |
449 450 |
* Ramp up sizes, and push forward the readahead window. */ |
08eb9658a mm: rename variou... |
451 452 |
if ((index == (ra->start + ra->size - ra->async_size) || index == (ra->start + ra->size))) { |
f9acc8c7b readahead: sanify... |
453 |
ra->start += ra->size; |
9491ae4aa mm: don't cap req... |
454 |
ra->size = get_next_ra_size(ra, max_pages); |
f9acc8c7b readahead: sanify... |
455 456 |
ra->async_size = ra->size; goto readit; |
122a21d11 readahead: on-dem... |
457 |
} |
122a21d11 readahead: on-dem... |
458 |
/* |
6b10c6c9f readahead: basic ... |
459 460 461 462 463 464 465 |
* Hit a marked page without valid readahead state. * E.g. interleaved reads. * Query the pagecache for async_size, which normally equals to * readahead size. Ramp it up and use it as the new readahead size. */ if (hit_readahead_marker) { pgoff_t start; |
30002ed2e mm: readahead sca... |
466 |
rcu_read_lock(); |
6e4af69ae mm/readahead: mak... |
467 468 |
start = page_cache_next_miss(ractl->mapping, index + 1, max_pages); |
30002ed2e mm: readahead sca... |
469 |
rcu_read_unlock(); |
6b10c6c9f readahead: basic ... |
470 |
|
08eb9658a mm: rename variou... |
471 |
if (!start || start - index > max_pages) |
9a42823a1 mm: return void f... |
472 |
return; |
6b10c6c9f readahead: basic ... |
473 474 |
ra->start = start; |
08eb9658a mm: rename variou... |
475 |
ra->size = start - index; /* old async_size */ |
160334a0c readahead: increa... |
476 |
ra->size += req_size; |
9491ae4aa mm: don't cap req... |
477 |
ra->size = get_next_ra_size(ra, max_pages); |
6b10c6c9f readahead: basic ... |
478 479 480 481 482 |
ra->async_size = ra->size; goto readit; } /* |
045a2529a readahead: move t... |
483 |
* oversize read |
122a21d11 readahead: on-dem... |
484 |
*/ |
9491ae4aa mm: don't cap req... |
485 |
if (req_size > max_pages) |
045a2529a readahead: move t... |
486 487 488 489 |
goto initial_readahead; /* * sequential cache miss |
08eb9658a mm: rename variou... |
490 491 |
* trivial case: (index - prev_index) == 1 * unaligned reads: (index - prev_index) == 0 |
045a2529a readahead: move t... |
492 |
*/ |
08eb9658a mm: rename variou... |
493 494 |
prev_index = (unsigned long long)ra->prev_pos >> PAGE_SHIFT; if (index - prev_index <= 1UL) |
045a2529a readahead: move t... |
495 496 497 |
goto initial_readahead; /* |
10be0b372 readahead: introd... |
498 499 500 |
* Query the page cache and look for the traces(cached history pages) * that a sequential stream would leave behind. */ |
6e4af69ae mm/readahead: mak... |
501 502 |
if (try_context_readahead(ractl->mapping, ra, index, req_size, max_pages)) |
10be0b372 readahead: introd... |
503 504 505 |
goto readit; /* |
045a2529a readahead: move t... |
506 507 508 |
* standalone, small random read * Read as is, and do not pollute the readahead state. */ |
6e4af69ae mm/readahead: mak... |
509 |
do_page_cache_ra(ractl, req_size, 0); |
9a42823a1 mm: return void f... |
510 |
return; |
045a2529a readahead: move t... |
511 512 |
initial_readahead: |
08eb9658a mm: rename variou... |
513 |
ra->start = index; |
9491ae4aa mm: don't cap req... |
514 |
ra->size = get_init_ra_size(req_size, max_pages); |
f9acc8c7b readahead: sanify... |
515 |
ra->async_size = ra->size > req_size ? ra->size - req_size : ra->size; |
122a21d11 readahead: on-dem... |
516 |
|
f9acc8c7b readahead: sanify... |
517 |
readit: |
51daa88eb readahead: remove... |
518 519 520 521 |
/* * Will this read hit the readahead marker made by itself? * If so, trigger the readahead marker hit now, and merge * the resulted next readahead window into the current one. |
dc30b96ab readahead: strict... |
522 |
* Take care of maximum IO pages as above. |
51daa88eb readahead: remove... |
523 |
*/ |
08eb9658a mm: rename variou... |
524 |
if (index == ra->start && ra->size == ra->async_size) { |
dc30b96ab readahead: strict... |
525 526 527 528 529 530 531 532 |
add_pages = get_next_ra_size(ra, max_pages); if (ra->size + add_pages <= max_pages) { ra->async_size = add_pages; ra->size += add_pages; } else { ra->size = max_pages; ra->async_size = max_pages >> 1; } |
51daa88eb readahead: remove... |
533 |
} |
6e4af69ae mm/readahead: mak... |
534 535 |
ractl->_index = ra->start; do_page_cache_ra(ractl, ra->size, ra->async_size); |
122a21d11 readahead: on-dem... |
536 |
} |
fefa7c478 mm/readahead: add... |
537 538 |
void page_cache_sync_ra(struct readahead_control *ractl, struct file_ra_state *ra, unsigned long req_count) |
122a21d11 readahead: on-dem... |
539 |
{ |
324bcf54c mm: use limited r... |
540 |
bool do_forced_ra = ractl->file && (ractl->file->f_mode & FMODE_RANDOM); |
cf914a7d6 readahead: split ... |
541 |
|
324bcf54c mm: use limited r... |
542 543 544 545 546 547 548 549 550 551 552 553 |
/* * Even if read-ahead is disabled, issue this request as read-ahead * as we'll need it to satisfy the requested range. The forced * read-ahead will do the right thing and limit the read to just the * requested range, which we'll set to 1 page for this case. */ if (!ra->ra_pages || blk_cgroup_congested()) { if (!ractl->file) return; req_count = 1; do_forced_ra = true; } |
ca47e8c72 mm: skip readahea... |
554 |
|
0141450f6 readahead: introd... |
555 |
/* be dumb */ |
324bcf54c mm: use limited r... |
556 |
if (do_forced_ra) { |
b1647dc0d mm/readahead: pas... |
557 |
force_page_cache_ra(ractl, ra, req_count); |
0141450f6 readahead: introd... |
558 559 |
return; } |
cf914a7d6 readahead: split ... |
560 |
/* do read-ahead */ |
fefa7c478 mm/readahead: add... |
561 |
ondemand_readahead(ractl, ra, false, req_count); |
cf914a7d6 readahead: split ... |
562 |
} |
fefa7c478 mm/readahead: add... |
563 |
EXPORT_SYMBOL_GPL(page_cache_sync_ra); |
cf914a7d6 readahead: split ... |
564 |
|
fefa7c478 mm/readahead: add... |
565 566 567 |
void page_cache_async_ra(struct readahead_control *ractl, struct file_ra_state *ra, struct page *page, unsigned long req_count) |
cf914a7d6 readahead: split ... |
568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 |
{ /* no read-ahead */ if (!ra->ra_pages) return; /* * Same bit is used for PG_readahead and PG_reclaim. */ if (PageWriteback(page)) return; ClearPageReadahead(page); /* * Defer asynchronous read-ahead on IO congestion. */ |
fefa7c478 mm/readahead: add... |
584 |
if (inode_read_congested(ractl->mapping->host)) |
cf914a7d6 readahead: split ... |
585 |
return; |
122a21d11 readahead: on-dem... |
586 |
|
ca47e8c72 mm: skip readahea... |
587 588 |
if (blk_cgroup_congested()) return; |
122a21d11 readahead: on-dem... |
589 |
/* do read-ahead */ |
fefa7c478 mm/readahead: add... |
590 |
ondemand_readahead(ractl, ra, true, req_count); |
122a21d11 readahead: on-dem... |
591 |
} |
fefa7c478 mm/readahead: add... |
592 |
EXPORT_SYMBOL_GPL(page_cache_async_ra); |
782182e53 mm: move readahea... |
593 |
|
c7b95d515 mm: add ksys_read... |
594 |
ssize_t ksys_readahead(int fd, loff_t offset, size_t count) |
782182e53 mm: move readahea... |
595 596 |
{ ssize_t ret; |
2903ff019 switch simple cas... |
597 |
struct fd f; |
782182e53 mm: move readahea... |
598 599 |
ret = -EBADF; |
2903ff019 switch simple cas... |
600 |
f = fdget(fd); |
3d8f76153 vfs: implement re... |
601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 |
if (!f.file || !(f.file->f_mode & FMODE_READ)) goto out; /* * The readahead() syscall is intended to run only on files * that can execute readahead. If readahead is not possible * on this file, then we must return -EINVAL. */ ret = -EINVAL; if (!f.file->f_mapping || !f.file->f_mapping->a_ops || !S_ISREG(file_inode(f.file)->i_mode)) goto out; ret = vfs_fadvise(f.file, offset, count, POSIX_FADV_WILLNEED); out: fdput(f); |
782182e53 mm: move readahea... |
617 618 |
return ret; } |
c7b95d515 mm: add ksys_read... |
619 620 621 622 623 |
SYSCALL_DEFINE3(readahead, int, fd, loff_t, offset, size_t, count) { return ksys_readahead(fd, offset, count); } |