Blame view
mm/readahead.c
16.5 KB
1da177e4c Linux-2.6.12-rc2 |
1 2 3 4 5 |
/* * mm/readahead.c - address_space-level file readahead. * * Copyright (C) 2002, Linus Torvalds * |
e1f8e8744 Remove Andrew Mor... |
6 |
* 09Apr2002 Andrew Morton |
1da177e4c Linux-2.6.12-rc2 |
7 8 9 10 |
* Initial version. */ #include <linux/kernel.h> |
11bd969fd mm: silently skip... |
11 |
#include <linux/dax.h> |
5a0e3ad6a include cleanup: ... |
12 |
#include <linux/gfp.h> |
b95f1b31b mm: Map most file... |
13 |
#include <linux/export.h> |
1da177e4c Linux-2.6.12-rc2 |
14 15 |
#include <linux/blkdev.h> #include <linux/backing-dev.h> |
8bde37f08 [PATCH] io-accoun... |
16 |
#include <linux/task_io_accounting_ops.h> |
1da177e4c Linux-2.6.12-rc2 |
17 |
#include <linux/pagevec.h> |
f5ff8422b Fix warnings with... |
18 |
#include <linux/pagemap.h> |
782182e53 mm: move readahea... |
19 20 |
#include <linux/syscalls.h> #include <linux/file.h> |
d72ee9111 mm: move lru_to_p... |
21 |
#include <linux/mm_inline.h> |
ca47e8c72 mm: skip readahea... |
22 |
#include <linux/blk-cgroup.h> |
3d8f76153 vfs: implement re... |
23 |
#include <linux/fadvise.h> |
1da177e4c Linux-2.6.12-rc2 |
24 |
|
29f175d12 mm/readahead.c: i... |
25 |
#include "internal.h" |
1da177e4c Linux-2.6.12-rc2 |
26 27 28 29 30 31 32 |
/* * 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_... |
33 |
ra->ra_pages = inode_to_bdi(mapping->host)->ra_pages; |
f4e6b498d readahead: combin... |
34 |
ra->prev_pos = -1; |
1da177e4c Linux-2.6.12-rc2 |
35 |
} |
d41cc702c [GFS2] Export fil... |
36 |
EXPORT_SYMBOL_GPL(file_ra_state_init); |
1da177e4c Linux-2.6.12-rc2 |
37 |
|
03fb3d2af FS-Cache: Release... |
38 39 |
/* * see if a page needs releasing upon read_cache_pages() failure |
266cf658e FS-Cache: Recruit... |
40 41 42 43 |
* - 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... |
44 45 46 47 |
*/ static void read_cache_pages_invalidate_page(struct address_space *mapping, struct page *page) { |
266cf658e FS-Cache: Recruit... |
48 |
if (page_has_private(page)) { |
03fb3d2af FS-Cache: Release... |
49 50 51 |
if (!trylock_page(page)) BUG(); page->mapping = mapping; |
09cbfeaf1 mm, fs: get rid o... |
52 |
do_invalidatepage(page, 0, PAGE_SIZE); |
03fb3d2af FS-Cache: Release... |
53 54 55 |
page->mapping = NULL; unlock_page(page); } |
09cbfeaf1 mm, fs: get rid o... |
56 |
put_page(page); |
03fb3d2af FS-Cache: Release... |
57 58 59 60 61 62 63 64 65 66 67 |
} /* * 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... |
68 |
victim = lru_to_page(pages); |
03fb3d2af FS-Cache: Release... |
69 70 71 72 |
list_del(&victim->lru); read_cache_pages_invalidate_page(mapping, victim); } } |
1da177e4c Linux-2.6.12-rc2 |
73 |
/** |
bd40cddae [PATCH] kernel-do... |
74 |
* read_cache_pages - populate an address space with some pages & start reads against them |
1da177e4c Linux-2.6.12-rc2 |
75 76 77 78 79 80 81 82 83 84 85 86 |
* @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. */ 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 |
87 |
int ret = 0; |
1da177e4c Linux-2.6.12-rc2 |
88 |
while (!list_empty(pages)) { |
c8ad6302c mm/readahead.c, m... |
89 |
page = lru_to_page(pages); |
1da177e4c Linux-2.6.12-rc2 |
90 |
list_del(&page->lru); |
063d99b4f mm, fs: obey gfp_... |
91 |
if (add_to_page_cache_lru(page, mapping, page->index, |
8a5c743e3 mm, memcg: use co... |
92 |
readahead_gfp_mask(mapping))) { |
03fb3d2af FS-Cache: Release... |
93 |
read_cache_pages_invalidate_page(mapping, page); |
1da177e4c Linux-2.6.12-rc2 |
94 95 |
continue; } |
09cbfeaf1 mm, fs: get rid o... |
96 |
put_page(page); |
eb2be1893 mm: buffered writ... |
97 |
|
1da177e4c Linux-2.6.12-rc2 |
98 |
ret = filler(data, page); |
eb2be1893 mm: buffered writ... |
99 |
if (unlikely(ret)) { |
03fb3d2af FS-Cache: Release... |
100 |
read_cache_pages_invalidate_pages(mapping, pages); |
1da177e4c Linux-2.6.12-rc2 |
101 102 |
break; } |
09cbfeaf1 mm, fs: get rid o... |
103 |
task_io_account_read(PAGE_SIZE); |
1da177e4c Linux-2.6.12-rc2 |
104 |
} |
1da177e4c Linux-2.6.12-rc2 |
105 106 107 108 109 110 |
return ret; } EXPORT_SYMBOL(read_cache_pages); static int read_pages(struct address_space *mapping, struct file *filp, |
8a5c743e3 mm, memcg: use co... |
111 |
struct list_head *pages, unsigned int nr_pages, gfp_t gfp) |
1da177e4c Linux-2.6.12-rc2 |
112 |
{ |
5b417b187 read-ahead: use p... |
113 |
struct blk_plug plug; |
1da177e4c Linux-2.6.12-rc2 |
114 |
unsigned page_idx; |
994fc28c7 [PATCH] add AOP_T... |
115 |
int ret; |
1da177e4c Linux-2.6.12-rc2 |
116 |
|
5b417b187 read-ahead: use p... |
117 |
blk_start_plug(&plug); |
1da177e4c Linux-2.6.12-rc2 |
118 119 |
if (mapping->a_ops->readpages) { ret = mapping->a_ops->readpages(filp, mapping, pages, nr_pages); |
029e332ea [PATCH] Cleanup r... |
120 121 |
/* Clean up the remaining pages */ put_pages_list(pages); |
1da177e4c Linux-2.6.12-rc2 |
122 123 |
goto out; } |
1da177e4c Linux-2.6.12-rc2 |
124 |
for (page_idx = 0; page_idx < nr_pages; page_idx++) { |
c8ad6302c mm/readahead.c, m... |
125 |
struct page *page = lru_to_page(pages); |
1da177e4c Linux-2.6.12-rc2 |
126 |
list_del(&page->lru); |
8a5c743e3 mm, memcg: use co... |
127 |
if (!add_to_page_cache_lru(page, mapping, page->index, gfp)) |
9f1a3cfcf [PATCH] AOP_TRUNC... |
128 |
mapping->a_ops->readpage(filp, page); |
09cbfeaf1 mm, fs: get rid o... |
129 |
put_page(page); |
1da177e4c Linux-2.6.12-rc2 |
130 |
} |
994fc28c7 [PATCH] add AOP_T... |
131 |
ret = 0; |
5b417b187 read-ahead: use p... |
132 |
|
1da177e4c Linux-2.6.12-rc2 |
133 |
out: |
5b417b187 read-ahead: use p... |
134 |
blk_finish_plug(&plug); |
1da177e4c Linux-2.6.12-rc2 |
135 136 137 138 |
return ret; } /* |
b3751e6ab mm: split ->readp... |
139 140 |
* __do_page_cache_readahead() actually reads a chunk of disk. It allocates * the pages first, then submits them for I/O. This avoids the very bad |
1da177e4c Linux-2.6.12-rc2 |
141 142 143 144 |
* behaviour which would occur if page allocations are causing VM writeback. * We really don't want to intermingle reads and writes like that. * * Returns the number of pages requested, or the maximum amount of I/O allowed. |
1da177e4c Linux-2.6.12-rc2 |
145 |
*/ |
c534aa3fd mm: return an uns... |
146 147 148 |
unsigned int __do_page_cache_readahead(struct address_space *mapping, struct file *filp, pgoff_t offset, unsigned long nr_to_read, unsigned long lookahead_size) |
1da177e4c Linux-2.6.12-rc2 |
149 150 151 152 153 154 |
{ struct inode *inode = mapping->host; struct page *page; unsigned long end_index; /* The last page we want to read */ LIST_HEAD(page_pool); int page_idx; |
c534aa3fd mm: return an uns... |
155 |
unsigned int nr_pages = 0; |
1da177e4c Linux-2.6.12-rc2 |
156 |
loff_t isize = i_size_read(inode); |
8a5c743e3 mm, memcg: use co... |
157 |
gfp_t gfp_mask = readahead_gfp_mask(mapping); |
1da177e4c Linux-2.6.12-rc2 |
158 159 160 |
if (isize == 0) goto out; |
09cbfeaf1 mm, fs: get rid o... |
161 |
end_index = ((isize - 1) >> PAGE_SHIFT); |
1da177e4c Linux-2.6.12-rc2 |
162 163 164 165 |
/* * Preallocate as many pages as we will need. */ |
1da177e4c Linux-2.6.12-rc2 |
166 |
for (page_idx = 0; page_idx < nr_to_read; page_idx++) { |
7361f4d8c [PATCH] readahead... |
167 |
pgoff_t page_offset = offset + page_idx; |
c743d96b6 readahead: remove... |
168 |
|
1da177e4c Linux-2.6.12-rc2 |
169 170 |
if (page_offset > end_index) break; |
001281881 mm: use lockless ... |
171 |
rcu_read_lock(); |
b93b01631 page cache: use x... |
172 |
page = radix_tree_lookup(&mapping->i_pages, page_offset); |
001281881 mm: use lockless ... |
173 |
rcu_read_unlock(); |
b3751e6ab mm: split ->readp... |
174 175 176 177 178 179 180 181 182 183 |
if (page && !radix_tree_exceptional_entry(page)) { /* * Page already present? Kick off the current batch of * contiguous pages before continuing with the next * batch. */ if (nr_pages) read_pages(mapping, filp, &page_pool, nr_pages, gfp_mask); nr_pages = 0; |
1da177e4c Linux-2.6.12-rc2 |
184 |
continue; |
b3751e6ab mm: split ->readp... |
185 |
} |
1da177e4c Linux-2.6.12-rc2 |
186 |
|
8a5c743e3 mm, memcg: use co... |
187 |
page = __page_cache_alloc(gfp_mask); |
1da177e4c Linux-2.6.12-rc2 |
188 189 190 191 |
if (!page) break; page->index = page_offset; list_add(&page->lru, &page_pool); |
46fc3e7b4 readahead: add lo... |
192 193 |
if (page_idx == nr_to_read - lookahead_size) SetPageReadahead(page); |
836978b35 mm: give the 'ret... |
194 |
nr_pages++; |
1da177e4c Linux-2.6.12-rc2 |
195 |
} |
1da177e4c Linux-2.6.12-rc2 |
196 197 198 199 200 201 |
/* * 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. */ |
836978b35 mm: give the 'ret... |
202 203 |
if (nr_pages) read_pages(mapping, filp, &page_pool, nr_pages, gfp_mask); |
1da177e4c Linux-2.6.12-rc2 |
204 205 |
BUG_ON(!list_empty(&page_pool)); out: |
836978b35 mm: give the 'ret... |
206 |
return nr_pages; |
1da177e4c Linux-2.6.12-rc2 |
207 208 209 210 211 212 213 |
} /* * Chunk the readahead into 2 megabyte units, so that we don't pin too much * memory at once. */ int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
9491ae4aa mm: don't cap req... |
214 |
pgoff_t offset, unsigned long nr_to_read) |
1da177e4c Linux-2.6.12-rc2 |
215 |
{ |
9491ae4aa mm: don't cap req... |
216 217 218 |
struct backing_dev_info *bdi = inode_to_bdi(mapping->host); struct file_ra_state *ra = &filp->f_ra; unsigned long max_pages; |
1da177e4c Linux-2.6.12-rc2 |
219 220 |
if (unlikely(!mapping->a_ops->readpage && !mapping->a_ops->readpages)) return -EINVAL; |
9491ae4aa mm: don't cap req... |
221 222 223 224 225 226 |
/* * If the request exceeds the readahead window, allow the read to * be up to the optimal hardware IO size */ max_pages = max_t(unsigned long, bdi->io_pages, ra->ra_pages); nr_to_read = min(nr_to_read, max_pages); |
1da177e4c Linux-2.6.12-rc2 |
227 |
while (nr_to_read) { |
09cbfeaf1 mm, fs: get rid o... |
228 |
unsigned long this_chunk = (2 * 1024 * 1024) / PAGE_SIZE; |
1da177e4c Linux-2.6.12-rc2 |
229 230 231 |
if (this_chunk > nr_to_read) this_chunk = nr_to_read; |
c534aa3fd mm: return an uns... |
232 |
__do_page_cache_readahead(mapping, filp, offset, this_chunk, 0); |
58d5640eb mm/readahead.c: f... |
233 |
|
1da177e4c Linux-2.6.12-rc2 |
234 235 236 |
offset += this_chunk; nr_to_read -= this_chunk; } |
58d5640eb mm/readahead.c: f... |
237 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
238 |
} |
5ce1110b9 readahead: data s... |
239 |
/* |
c743d96b6 readahead: remove... |
240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 |
* 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... |
260 261 262 |
* Get the previous window size, ramp it up, and * return it as the new window size. */ |
c743d96b6 readahead: remove... |
263 |
static unsigned long get_next_ra_size(struct file_ra_state *ra, |
122a21d11 readahead: on-dem... |
264 265 |
unsigned long max) { |
f9acc8c7b readahead: sanify... |
266 |
unsigned long cur = ra->size; |
122a21d11 readahead: on-dem... |
267 268 269 |
unsigned long newsize; if (cur < max / 16) |
c743d96b6 readahead: remove... |
270 |
newsize = 4 * cur; |
122a21d11 readahead: on-dem... |
271 |
else |
c743d96b6 readahead: remove... |
272 |
newsize = 2 * cur; |
122a21d11 readahead: on-dem... |
273 274 275 276 277 278 279 280 281 282 |
return min(newsize, max); } /* * On-demand readahead design. * * The fields in struct file_ra_state represent the most-recently-executed * readahead attempt: * |
f9acc8c7b readahead: sanify... |
283 284 285 286 |
* |<----- async_size ---------| * |------------------- size -------------------->| * |==================#===========================| * ^start ^page marked with PG_readahead |
122a21d11 readahead: on-dem... |
287 288 289 290 |
* * 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... |
291 292 293 |
* 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... |
294 295 296 |
* * 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... |
297 |
* page at (start+size-async_size) with PG_readahead, and use it as readahead |
122a21d11 readahead: on-dem... |
298 299 300 |
* 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... |
301 |
* prev_pos tracks the last visited byte in the _previous_ read request. |
122a21d11 readahead: on-dem... |
302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 |
* 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. */ /* |
10be0b372 readahead: introd... |
317 318 319 320 321 322 |
* Count contiguously cached pages from @offset-1 to @offset-@max, * 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, |
10be0b372 readahead: introd... |
323 324 325 326 327 |
pgoff_t offset, unsigned long max) { pgoff_t head; rcu_read_lock(); |
e7b563bb2 mm: filemap: move... |
328 |
head = page_cache_prev_hole(mapping, offset - 1, max); |
10be0b372 readahead: introd... |
329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 |
rcu_read_unlock(); return offset - 1 - head; } /* * page cache context based read-ahead */ static int try_context_readahead(struct address_space *mapping, struct file_ra_state *ra, pgoff_t offset, unsigned long req_size, unsigned long max) { pgoff_t size; |
3e2faa085 mm/readahead.c: r... |
344 |
size = count_history_pages(mapping, offset, max); |
10be0b372 readahead: introd... |
345 346 |
/* |
2cad40180 readahead: make c... |
347 |
* not enough history pages: |
10be0b372 readahead: introd... |
348 349 |
* it could be a random read */ |
2cad40180 readahead: make c... |
350 |
if (size <= req_size) |
10be0b372 readahead: introd... |
351 352 353 354 355 356 357 358 359 360 |
return 0; /* * starts from beginning of file: * it is a strong indication of long-run stream (or whole-file-read) */ if (size >= offset) size *= 2; ra->start = offset; |
2cad40180 readahead: make c... |
361 362 |
ra->size = min(size + req_size, max); ra->async_size = 1; |
10be0b372 readahead: introd... |
363 364 365 366 367 |
return 1; } /* |
122a21d11 readahead: on-dem... |
368 369 370 371 372 |
* A minimal readahead algorithm for trivial sequential/random reads. */ static unsigned long ondemand_readahead(struct address_space *mapping, struct file_ra_state *ra, struct file *filp, |
cf914a7d6 readahead: split ... |
373 |
bool hit_readahead_marker, pgoff_t offset, |
122a21d11 readahead: on-dem... |
374 375 |
unsigned long req_size) { |
9491ae4aa mm: don't cap req... |
376 377 |
struct backing_dev_info *bdi = inode_to_bdi(mapping->host); unsigned long max_pages = ra->ra_pages; |
dc30b96ab readahead: strict... |
378 |
unsigned long add_pages; |
af248a0c6 readahead: fix se... |
379 |
pgoff_t prev_offset; |
045a2529a readahead: move t... |
380 381 |
/* |
9491ae4aa mm: don't cap req... |
382 383 384 385 386 387 388 |
* 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... |
389 390 391 392 |
* start of file */ if (!offset) goto initial_readahead; |
122a21d11 readahead: on-dem... |
393 394 |
/* |
f9acc8c7b readahead: sanify... |
395 |
* It's the expected callback offset, assume sequential access. |
122a21d11 readahead: on-dem... |
396 397 |
* Ramp up sizes, and push forward the readahead window. */ |
045a2529a readahead: move t... |
398 399 |
if ((offset == (ra->start + ra->size - ra->async_size) || offset == (ra->start + ra->size))) { |
f9acc8c7b readahead: sanify... |
400 |
ra->start += ra->size; |
9491ae4aa mm: don't cap req... |
401 |
ra->size = get_next_ra_size(ra, max_pages); |
f9acc8c7b readahead: sanify... |
402 403 |
ra->async_size = ra->size; goto readit; |
122a21d11 readahead: on-dem... |
404 |
} |
122a21d11 readahead: on-dem... |
405 |
/* |
6b10c6c9f readahead: basic ... |
406 407 408 409 410 411 412 |
* 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... |
413 |
rcu_read_lock(); |
9491ae4aa mm: don't cap req... |
414 |
start = page_cache_next_hole(mapping, offset + 1, max_pages); |
30002ed2e mm: readahead sca... |
415 |
rcu_read_unlock(); |
6b10c6c9f readahead: basic ... |
416 |
|
9491ae4aa mm: don't cap req... |
417 |
if (!start || start - offset > max_pages) |
6b10c6c9f readahead: basic ... |
418 419 420 421 |
return 0; ra->start = start; ra->size = start - offset; /* old async_size */ |
160334a0c readahead: increa... |
422 |
ra->size += req_size; |
9491ae4aa mm: don't cap req... |
423 |
ra->size = get_next_ra_size(ra, max_pages); |
6b10c6c9f readahead: basic ... |
424 425 426 427 428 |
ra->async_size = ra->size; goto readit; } /* |
045a2529a readahead: move t... |
429 |
* oversize read |
122a21d11 readahead: on-dem... |
430 |
*/ |
9491ae4aa mm: don't cap req... |
431 |
if (req_size > max_pages) |
045a2529a readahead: move t... |
432 433 434 435 |
goto initial_readahead; /* * sequential cache miss |
af248a0c6 readahead: fix se... |
436 437 |
* trivial case: (offset - prev_offset) == 1 * unaligned reads: (offset - prev_offset) == 0 |
045a2529a readahead: move t... |
438 |
*/ |
09cbfeaf1 mm, fs: get rid o... |
439 |
prev_offset = (unsigned long long)ra->prev_pos >> PAGE_SHIFT; |
af248a0c6 readahead: fix se... |
440 |
if (offset - prev_offset <= 1UL) |
045a2529a readahead: move t... |
441 442 443 |
goto initial_readahead; /* |
10be0b372 readahead: introd... |
444 445 446 |
* Query the page cache and look for the traces(cached history pages) * that a sequential stream would leave behind. */ |
9491ae4aa mm: don't cap req... |
447 |
if (try_context_readahead(mapping, ra, offset, req_size, max_pages)) |
10be0b372 readahead: introd... |
448 449 450 |
goto readit; /* |
045a2529a readahead: move t... |
451 452 453 454 455 456 |
* standalone, small random read * Read as is, and do not pollute the readahead state. */ return __do_page_cache_readahead(mapping, filp, offset, req_size, 0); initial_readahead: |
f9acc8c7b readahead: sanify... |
457 |
ra->start = offset; |
9491ae4aa mm: don't cap req... |
458 |
ra->size = get_init_ra_size(req_size, max_pages); |
f9acc8c7b readahead: sanify... |
459 |
ra->async_size = ra->size > req_size ? ra->size - req_size : ra->size; |
122a21d11 readahead: on-dem... |
460 |
|
f9acc8c7b readahead: sanify... |
461 |
readit: |
51daa88eb readahead: remove... |
462 463 464 465 |
/* * 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... |
466 |
* Take care of maximum IO pages as above. |
51daa88eb readahead: remove... |
467 468 |
*/ if (offset == ra->start && ra->size == ra->async_size) { |
dc30b96ab readahead: strict... |
469 470 471 472 473 474 475 476 |
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... |
477 |
} |
122a21d11 readahead: on-dem... |
478 479 480 481 |
return ra_submit(ra, mapping, filp); } /** |
cf914a7d6 readahead: split ... |
482 |
* page_cache_sync_readahead - generic file readahead |
122a21d11 readahead: on-dem... |
483 484 485 |
* @mapping: address_space which holds the pagecache and I/O vectors * @ra: file_ra_state which holds the readahead state * @filp: passed on to ->readpage() and ->readpages() |
cf914a7d6 readahead: split ... |
486 |
* @offset: start offset into @mapping, in pagecache page-sized units |
122a21d11 readahead: on-dem... |
487 |
* @req_size: hint: total size of the read which the caller is performing in |
cf914a7d6 readahead: split ... |
488 |
* pagecache pages |
122a21d11 readahead: on-dem... |
489 |
* |
cf914a7d6 readahead: split ... |
490 491 492 493 |
* page_cache_sync_readahead() should be called when a cache miss happened: * it will submit the read. The readahead logic may decide to piggyback more * pages onto the read request if access patterns suggest it will improve * performance. |
122a21d11 readahead: on-dem... |
494 |
*/ |
cf914a7d6 readahead: split ... |
495 496 497 |
void page_cache_sync_readahead(struct address_space *mapping, struct file_ra_state *ra, struct file *filp, pgoff_t offset, unsigned long req_size) |
122a21d11 readahead: on-dem... |
498 499 500 |
{ /* no read-ahead */ if (!ra->ra_pages) |
cf914a7d6 readahead: split ... |
501 |
return; |
ca47e8c72 mm: skip readahea... |
502 503 |
if (blk_cgroup_congested()) return; |
0141450f6 readahead: introd... |
504 |
/* be dumb */ |
70655c06b readahead: fix NU... |
505 |
if (filp && (filp->f_mode & FMODE_RANDOM)) { |
0141450f6 readahead: introd... |
506 507 508 |
force_page_cache_readahead(mapping, filp, offset, req_size); return; } |
cf914a7d6 readahead: split ... |
509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 |
/* do read-ahead */ ondemand_readahead(mapping, ra, filp, false, offset, req_size); } EXPORT_SYMBOL_GPL(page_cache_sync_readahead); /** * page_cache_async_readahead - file readahead for marked pages * @mapping: address_space which holds the pagecache and I/O vectors * @ra: file_ra_state which holds the readahead state * @filp: passed on to ->readpage() and ->readpages() * @page: the page at @offset which has the PG_readahead flag set * @offset: start offset into @mapping, in pagecache page-sized units * @req_size: hint: total size of the read which the caller is performing in * pagecache pages * |
bf8abe8b9 readahead.c: fix ... |
524 |
* page_cache_async_readahead() should be called when a page is used which |
f7850d932 mm/readahead: fix... |
525 |
* has the PG_readahead flag; this is a marker to suggest that the application |
cf914a7d6 readahead: split ... |
526 |
* has used up enough of the readahead window that we should start pulling in |
f7850d932 mm/readahead: fix... |
527 528 |
* more pages. */ |
cf914a7d6 readahead: split ... |
529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 |
void page_cache_async_readahead(struct address_space *mapping, struct file_ra_state *ra, struct file *filp, struct page *page, pgoff_t offset, unsigned long req_size) { /* 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. */ |
703c27088 writeback: implem... |
550 |
if (inode_read_congested(mapping->host)) |
cf914a7d6 readahead: split ... |
551 |
return; |
122a21d11 readahead: on-dem... |
552 |
|
ca47e8c72 mm: skip readahea... |
553 554 |
if (blk_cgroup_congested()) return; |
122a21d11 readahead: on-dem... |
555 |
/* do read-ahead */ |
cf914a7d6 readahead: split ... |
556 |
ondemand_readahead(mapping, ra, filp, true, offset, req_size); |
122a21d11 readahead: on-dem... |
557 |
} |
cf914a7d6 readahead: split ... |
558 |
EXPORT_SYMBOL_GPL(page_cache_async_readahead); |
782182e53 mm: move readahea... |
559 |
|
c7b95d515 mm: add ksys_read... |
560 |
ssize_t ksys_readahead(int fd, loff_t offset, size_t count) |
782182e53 mm: move readahea... |
561 562 |
{ ssize_t ret; |
2903ff019 switch simple cas... |
563 |
struct fd f; |
782182e53 mm: move readahea... |
564 565 |
ret = -EBADF; |
2903ff019 switch simple cas... |
566 |
f = fdget(fd); |
3d8f76153 vfs: implement re... |
567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 |
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... |
583 584 |
return ret; } |
c7b95d515 mm: add ksys_read... |
585 586 587 588 589 |
SYSCALL_DEFINE3(readahead, int, fd, loff_t, offset, size_t, count) { return ksys_readahead(fd, offset, count); } |