Blame view
fs/exofs/inode.c
36.4 KB
e80627191 exofs: file and f... |
1 2 |
/* * Copyright (C) 2005, 2006 |
27d2e1491 exofs: Remove IBM... |
3 |
* Avishay Traeger (avishay@gmail.com) |
e80627191 exofs: file and f... |
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 |
* Copyright (C) 2008, 2009 * Boaz Harrosh <bharrosh@panasas.com> * * Copyrights for code taken from ext2: * Copyright (C) 1992, 1993, 1994, 1995 * Remy Card (card@masi.ibp.fr) * Laboratoire MASI - Institut Blaise Pascal * Universite Pierre et Marie Curie (Paris VI) * from * linux/fs/minix/inode.c * Copyright (C) 1991, 1992 Linus Torvalds * * This file is part of exofs. * * exofs is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation. Since it is based on ext2, and the only * valid version of GPL for the Linux kernel is version 2, the only valid * version of GPL for exofs is version 2. * * exofs is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with exofs; if not, write to the Free Software * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ |
5a0e3ad6a include cleanup: ... |
33 |
#include <linux/slab.h> |
e80627191 exofs: file and f... |
34 35 |
#include "exofs.h" |
fe33cc1ee exofs: dbg-print ... |
36 |
#define EXOFS_DBGMSG2(M...) do {} while (0) |
5a51c0c7e ore/exofs: Define... |
37 |
enum {MAX_PAGES_KMALLOC = PAGE_SIZE / sizeof(struct page *), }; |
06886a5a3 exofs: Move all o... |
38 |
|
8ff660ab8 exofs: Rename rai... |
39 |
unsigned exofs_max_io_pages(struct ore_layout *layout, |
66cd6cad4 exofs: Override r... |
40 41 42 43 44 |
unsigned expected_pages) { unsigned pages = min_t(unsigned, expected_pages, MAX_PAGES_KMALLOC); /* TODO: easily support bio chaining */ |
5a51c0c7e ore/exofs: Define... |
45 |
pages = min_t(unsigned, pages, layout->max_io_length / PAGE_SIZE); |
66cd6cad4 exofs: Override r... |
46 47 |
return pages; } |
beaec07ba exofs: address_sp... |
48 49 |
struct page_collect { struct exofs_sb_info *sbi; |
beaec07ba exofs: address_sp... |
50 51 |
struct inode *inode; unsigned expected_pages; |
8ff660ab8 exofs: Rename rai... |
52 |
struct ore_io_state *ios; |
beaec07ba exofs: address_sp... |
53 |
|
86093aaff exofs: convert io... |
54 55 |
struct page **pages; unsigned alloc_pages; |
beaec07ba exofs: address_sp... |
56 57 58 |
unsigned nr_pages; unsigned long length; loff_t pg_first; /* keep 64bit also in 32-arches */ |
f17b1f9f1 exofs: Fix double... |
59 60 61 |
bool read_4_write; /* This means two things: that the read is sync * And the pages should not be unlocked. */ |
dd2966199 exofs: Support fo... |
62 |
struct page *that_locked_page; |
beaec07ba exofs: address_sp... |
63 64 65 |
}; static void _pcol_init(struct page_collect *pcol, unsigned expected_pages, |
06886a5a3 exofs: Move all o... |
66 |
struct inode *inode) |
beaec07ba exofs: address_sp... |
67 68 |
{ struct exofs_sb_info *sbi = inode->i_sb->s_fs_info; |
beaec07ba exofs: address_sp... |
69 70 |
pcol->sbi = sbi; |
beaec07ba exofs: address_sp... |
71 72 |
pcol->inode = inode; pcol->expected_pages = expected_pages; |
06886a5a3 exofs: Move all o... |
73 |
pcol->ios = NULL; |
86093aaff exofs: convert io... |
74 75 |
pcol->pages = NULL; pcol->alloc_pages = 0; |
beaec07ba exofs: address_sp... |
76 77 78 |
pcol->nr_pages = 0; pcol->length = 0; pcol->pg_first = -1; |
f17b1f9f1 exofs: Fix double... |
79 |
pcol->read_4_write = false; |
dd2966199 exofs: Support fo... |
80 |
pcol->that_locked_page = NULL; |
beaec07ba exofs: address_sp... |
81 82 83 84 85 |
} static void _pcol_reset(struct page_collect *pcol) { pcol->expected_pages -= min(pcol->nr_pages, pcol->expected_pages); |
86093aaff exofs: convert io... |
86 87 |
pcol->pages = NULL; pcol->alloc_pages = 0; |
beaec07ba exofs: address_sp... |
88 89 90 |
pcol->nr_pages = 0; pcol->length = 0; pcol->pg_first = -1; |
06886a5a3 exofs: Move all o... |
91 |
pcol->ios = NULL; |
dd2966199 exofs: Support fo... |
92 |
pcol->that_locked_page = NULL; |
beaec07ba exofs: address_sp... |
93 94 95 96 97 |
/* this is probably the end of the loop but in writes * it might not end here. don't be left with nothing */ if (!pcol->expected_pages) |
86093aaff exofs: convert io... |
98 |
pcol->expected_pages = MAX_PAGES_KMALLOC; |
beaec07ba exofs: address_sp... |
99 100 101 102 |
} static int pcol_try_alloc(struct page_collect *pcol) { |
66cd6cad4 exofs: Override r... |
103 |
unsigned pages; |
06886a5a3 exofs: Move all o... |
104 |
|
86093aaff exofs: convert io... |
105 |
/* TODO: easily support bio chaining */ |
66cd6cad4 exofs: Override r... |
106 |
pages = exofs_max_io_pages(&pcol->sbi->layout, pcol->expected_pages); |
86093aaff exofs: convert io... |
107 |
|
beaec07ba exofs: address_sp... |
108 |
for (; pages; pages >>= 1) { |
86093aaff exofs: convert io... |
109 110 111 112 |
pcol->pages = kmalloc(pages * sizeof(struct page *), GFP_KERNEL); if (likely(pcol->pages)) { pcol->alloc_pages = pages; |
beaec07ba exofs: address_sp... |
113 |
return 0; |
86093aaff exofs: convert io... |
114 |
} |
beaec07ba exofs: address_sp... |
115 |
} |
86093aaff exofs: convert io... |
116 117 |
EXOFS_ERR("Failed to kmalloc expected_pages=%u ", |
beaec07ba exofs: address_sp... |
118 119 120 121 122 123 |
pcol->expected_pages); return -ENOMEM; } static void pcol_free(struct page_collect *pcol) { |
86093aaff exofs: convert io... |
124 125 |
kfree(pcol->pages); pcol->pages = NULL; |
06886a5a3 exofs: Move all o... |
126 127 |
if (pcol->ios) { |
8ff660ab8 exofs: Rename rai... |
128 |
ore_put_io_state(pcol->ios); |
06886a5a3 exofs: Move all o... |
129 130 |
pcol->ios = NULL; } |
beaec07ba exofs: address_sp... |
131 132 133 134 135 |
} static int pcol_add_page(struct page_collect *pcol, struct page *page, unsigned len) { |
86093aaff exofs: convert io... |
136 |
if (unlikely(pcol->nr_pages >= pcol->alloc_pages)) |
beaec07ba exofs: address_sp... |
137 |
return -ENOMEM; |
86093aaff exofs: convert io... |
138 |
pcol->pages[pcol->nr_pages++] = page; |
beaec07ba exofs: address_sp... |
139 140 141 |
pcol->length += len; return 0; } |
154a9300c exofs: Support fo... |
142 |
enum {PAGE_WAS_NOT_IN_IO = 17}; |
beaec07ba exofs: address_sp... |
143 144 |
static int update_read_page(struct page *page, int ret) { |
154a9300c exofs: Support fo... |
145 146 |
switch (ret) { case 0: |
beaec07ba exofs: address_sp... |
147 148 149 150 |
/* Everything is OK */ SetPageUptodate(page); if (PageError(page)) ClearPageError(page); |
154a9300c exofs: Support fo... |
151 152 |
break; case -EFAULT: |
beaec07ba exofs: address_sp... |
153 154 155 156 157 158 159 160 161 162 |
/* In this case we were trying to read something that wasn't on * disk yet - return a page full of zeroes. This should be OK, * because the object should be empty (if there was a write * before this read, the read would be waiting with the page * locked */ clear_highpage(page); SetPageUptodate(page); if (PageError(page)) ClearPageError(page); |
beaec07ba exofs: address_sp... |
163 164 |
EXOFS_DBGMSG("recovered read error "); |
154a9300c exofs: Support fo... |
165 166 167 168 169 |
/* fall through */ case PAGE_WAS_NOT_IN_IO: ret = 0; /* recovered error */ break; default: |
beaec07ba exofs: address_sp... |
170 |
SetPageError(page); |
154a9300c exofs: Support fo... |
171 |
} |
beaec07ba exofs: address_sp... |
172 173 174 175 176 |
return ret; } static void update_write_page(struct page *page, int ret) { |
154a9300c exofs: Support fo... |
177 178 |
if (unlikely(ret == PAGE_WAS_NOT_IN_IO)) return; /* don't pass start don't collect $200 */ |
beaec07ba exofs: address_sp... |
179 180 181 182 183 184 185 186 187 188 |
if (ret) { mapping_set_error(page->mapping, ret); SetPageError(page); } end_page_writeback(page); } /* Called at the end of reads, to optionally unlock pages and update their * status. */ |
7aebf4106 exofs: Cleaup rea... |
189 |
static int __readpages_done(struct page_collect *pcol) |
beaec07ba exofs: address_sp... |
190 |
{ |
beaec07ba exofs: address_sp... |
191 |
int i; |
beaec07ba exofs: address_sp... |
192 193 |
u64 good_bytes; u64 length = 0; |
4b46c9f5c ore/exofs: Change... |
194 |
int ret = ore_check_io(pcol->ios, NULL); |
beaec07ba exofs: address_sp... |
195 |
|
154a9300c exofs: Support fo... |
196 |
if (likely(!ret)) { |
beaec07ba exofs: address_sp... |
197 |
good_bytes = pcol->length; |
154a9300c exofs: Support fo... |
198 199 |
ret = PAGE_WAS_NOT_IN_IO; } else { |
4b46c9f5c ore/exofs: Change... |
200 |
good_bytes = 0; |
154a9300c exofs: Support fo... |
201 |
} |
beaec07ba exofs: address_sp... |
202 |
|
34ce4e7c2 exofs: debug prin... |
203 |
EXOFS_DBGMSG2("readpages_done(0x%lx) good_bytes=0x%llx" |
beaec07ba exofs: address_sp... |
204 205 206 207 |
" length=0x%lx nr_pages=%u ", pcol->inode->i_ino, _LLU(good_bytes), pcol->length, pcol->nr_pages); |
86093aaff exofs: convert io... |
208 209 |
for (i = 0; i < pcol->nr_pages; i++) { struct page *page = pcol->pages[i]; |
beaec07ba exofs: address_sp... |
210 211 212 213 214 215 216 217 218 219 |
struct inode *inode = page->mapping->host; int page_stat; if (inode != pcol->inode) continue; /* osd might add more pages at end */ if (likely(length < good_bytes)) page_stat = 0; else page_stat = ret; |
fe33cc1ee exofs: dbg-print ... |
220 221 |
EXOFS_DBGMSG2(" readpages_done(0x%lx, 0x%lx) %s ", |
beaec07ba exofs: address_sp... |
222 223 224 225 |
inode->i_ino, page->index, page_stat ? "bad_bytes" : "good_bytes"); ret = update_read_page(page, page_stat); |
7aebf4106 exofs: Cleaup rea... |
226 |
if (!pcol->read_4_write) |
beaec07ba exofs: address_sp... |
227 |
unlock_page(page); |
86093aaff exofs: convert io... |
228 |
length += PAGE_SIZE; |
beaec07ba exofs: address_sp... |
229 230 231 |
} pcol_free(pcol); |
34ce4e7c2 exofs: debug prin... |
232 233 |
EXOFS_DBGMSG2("readpages_done END "); |
beaec07ba exofs: address_sp... |
234 235 236 237 |
return ret; } /* callback of async reads */ |
8ff660ab8 exofs: Rename rai... |
238 |
static void readpages_done(struct ore_io_state *ios, void *p) |
beaec07ba exofs: address_sp... |
239 240 |
{ struct page_collect *pcol = p; |
7aebf4106 exofs: Cleaup rea... |
241 |
__readpages_done(pcol); |
beaec07ba exofs: address_sp... |
242 |
atomic_dec(&pcol->sbi->s_curr_pending); |
06886a5a3 exofs: Move all o... |
243 |
kfree(pcol); |
beaec07ba exofs: address_sp... |
244 245 246 247 |
} static void _unlock_pcol_pages(struct page_collect *pcol, int ret, int rw) { |
beaec07ba exofs: address_sp... |
248 |
int i; |
86093aaff exofs: convert io... |
249 250 |
for (i = 0; i < pcol->nr_pages; i++) { struct page *page = pcol->pages[i]; |
beaec07ba exofs: address_sp... |
251 252 253 254 255 256 257 258 |
if (rw == READ) update_read_page(page, ret); else update_write_page(page, ret); unlock_page(page); } |
beaec07ba exofs: address_sp... |
259 |
} |
b916c5cd4 ore: Only IO one ... |
260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 |
static int _maybe_not_all_in_one_io(struct ore_io_state *ios, struct page_collect *pcol_src, struct page_collect *pcol) { /* length was wrong or offset was not page aligned */ BUG_ON(pcol_src->nr_pages < ios->nr_pages); if (pcol_src->nr_pages > ios->nr_pages) { struct page **src_page; unsigned pages_less = pcol_src->nr_pages - ios->nr_pages; unsigned long len_less = pcol_src->length - ios->length; unsigned i; int ret; /* This IO was trimmed */ pcol_src->nr_pages = ios->nr_pages; pcol_src->length = ios->length; /* Left over pages are passed to the next io */ pcol->expected_pages += pages_less; pcol->nr_pages = pages_less; pcol->length = len_less; src_page = pcol_src->pages + pcol_src->nr_pages; pcol->pg_first = (*src_page)->index; ret = pcol_try_alloc(pcol); if (unlikely(ret)) return ret; for (i = 0; i < pages_less; ++i) pcol->pages[i] = *src_page++; EXOFS_DBGMSG("Length was adjusted nr_pages=0x%x " "pages_less=0x%x expected_pages=0x%x " "next_offset=0x%llx next_len=0x%lx ", pcol_src->nr_pages, pages_less, pcol->expected_pages, pcol->pg_first * PAGE_SIZE, pcol->length); } return 0; } |
7aebf4106 exofs: Cleaup rea... |
300 |
static int read_exec(struct page_collect *pcol) |
beaec07ba exofs: address_sp... |
301 302 |
{ struct exofs_i_info *oi = exofs_i(pcol->inode); |
8ff660ab8 exofs: Rename rai... |
303 |
struct ore_io_state *ios; |
beaec07ba exofs: address_sp... |
304 |
struct page_collect *pcol_copy = NULL; |
beaec07ba exofs: address_sp... |
305 |
int ret; |
86093aaff exofs: convert io... |
306 |
if (!pcol->pages) |
beaec07ba exofs: address_sp... |
307 |
return 0; |
e1042ba09 exofs: Add offset... |
308 |
if (!pcol->ios) { |
5bf696dad exofs: Rename str... |
309 |
int ret = ore_get_rw_state(&pcol->sbi->layout, &oi->oc, true, |
e1042ba09 exofs: Add offset... |
310 311 312 313 314 315 316 317 |
pcol->pg_first << PAGE_CACHE_SHIFT, pcol->length, &pcol->ios); if (ret) return ret; } ios = pcol->ios; |
86093aaff exofs: convert io... |
318 |
ios->pages = pcol->pages; |
beaec07ba exofs: address_sp... |
319 |
|
7aebf4106 exofs: Cleaup rea... |
320 |
if (pcol->read_4_write) { |
8ff660ab8 exofs: Rename rai... |
321 |
ore_read(pcol->ios); |
7aebf4106 exofs: Cleaup rea... |
322 |
return __readpages_done(pcol); |
beaec07ba exofs: address_sp... |
323 324 325 326 327 328 329 330 331 |
} pcol_copy = kmalloc(sizeof(*pcol_copy), GFP_KERNEL); if (!pcol_copy) { ret = -ENOMEM; goto err; } *pcol_copy = *pcol; |
06886a5a3 exofs: Move all o... |
332 333 |
ios->done = readpages_done; ios->private = pcol_copy; |
b916c5cd4 ore: Only IO one ... |
334 335 336 337 338 339 340 341 342 343 344 |
/* pages ownership was passed to pcol_copy */ _pcol_reset(pcol); ret = _maybe_not_all_in_one_io(ios, pcol_copy, pcol); if (unlikely(ret)) goto err; EXOFS_DBGMSG2("read_exec(0x%lx) offset=0x%llx length=0x%llx ", pcol->inode->i_ino, _LLU(ios->offset), _LLU(ios->length)); |
8ff660ab8 exofs: Rename rai... |
345 |
ret = ore_read(ios); |
beaec07ba exofs: address_sp... |
346 347 348 349 |
if (unlikely(ret)) goto err; atomic_inc(&pcol->sbi->s_curr_pending); |
beaec07ba exofs: address_sp... |
350 351 352 |
return 0; err: |
7aebf4106 exofs: Cleaup rea... |
353 |
if (!pcol->read_4_write) |
beaec07ba exofs: address_sp... |
354 |
_unlock_pcol_pages(pcol, ret, READ); |
06886a5a3 exofs: Move all o... |
355 356 |
pcol_free(pcol); |
b76a3f93d exofs: Fix bio le... |
357 |
|
beaec07ba exofs: address_sp... |
358 |
kfree(pcol_copy); |
beaec07ba exofs: address_sp... |
359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 |
return ret; } /* readpage_strip is called either directly from readpage() or by the VFS from * within read_cache_pages(), to add one more page to be read. It will try to * collect as many contiguous pages as posible. If a discontinuity is * encountered, or it runs out of resources, it will submit the previous segment * and will start a new collection. Eventually caller must submit the last * segment if present. */ static int readpage_strip(void *data, struct page *page) { struct page_collect *pcol = data; struct inode *inode = pcol->inode; struct exofs_i_info *oi = exofs_i(inode); loff_t i_size = i_size_read(inode); pgoff_t end_index = i_size >> PAGE_CACHE_SHIFT; size_t len; int ret; /* FIXME: Just for debugging, will be removed */ if (PageUptodate(page)) EXOFS_ERR("PageUptodate(0x%lx, 0x%lx) ", pcol->inode->i_ino, page->index); |
dd2966199 exofs: Support fo... |
384 |
pcol->that_locked_page = page; |
beaec07ba exofs: address_sp... |
385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 |
if (page->index < end_index) len = PAGE_CACHE_SIZE; else if (page->index == end_index) len = i_size & ~PAGE_CACHE_MASK; else len = 0; if (!len || !obj_created(oi)) { /* this will be out of bounds, or doesn't exist yet. * Current page is cleared and the request is split */ clear_highpage(page); SetPageUptodate(page); if (PageError(page)) ClearPageError(page); |
f17b1f9f1 exofs: Fix double... |
401 402 |
if (!pcol->read_4_write) unlock_page(page); |
a8f1418f9 exofs: Optimize r... |
403 404 405 406 407 |
EXOFS_DBGMSG("readpage_strip(0x%lx) empty page len=%zx " "read_4_write=%d index=0x%lx end_index=0x%lx " "splitting ", inode->i_ino, len, pcol->read_4_write, page->index, end_index); |
beaec07ba exofs: address_sp... |
408 |
|
7aebf4106 exofs: Cleaup rea... |
409 |
return read_exec(pcol); |
beaec07ba exofs: address_sp... |
410 411 412 413 414 415 416 417 418 |
} try_again: if (unlikely(pcol->pg_first == -1)) { pcol->pg_first = page->index; } else if (unlikely((pcol->pg_first + pcol->nr_pages) != page->index)) { /* Discontinuity detected, split the request */ |
7aebf4106 exofs: Cleaup rea... |
419 |
ret = read_exec(pcol); |
beaec07ba exofs: address_sp... |
420 421 422 423 |
if (unlikely(ret)) goto fail; goto try_again; } |
86093aaff exofs: convert io... |
424 |
if (!pcol->pages) { |
beaec07ba exofs: address_sp... |
425 426 427 428 429 430 431 |
ret = pcol_try_alloc(pcol); if (unlikely(ret)) goto fail; } if (len != PAGE_CACHE_SIZE) zero_user(page, len, PAGE_CACHE_SIZE - len); |
fe33cc1ee exofs: dbg-print ... |
432 433 |
EXOFS_DBGMSG2(" readpage_strip(0x%lx, 0x%lx) len=0x%zx ", |
beaec07ba exofs: address_sp... |
434 435 436 437 |
inode->i_ino, page->index, len); ret = pcol_add_page(pcol, page, len); if (ret) { |
fe33cc1ee exofs: dbg-print ... |
438 |
EXOFS_DBGMSG2("Failed pcol_add_page pages[i]=%p " |
beaec07ba exofs: address_sp... |
439 440 441 442 443 |
"this_len=0x%zx nr_pages=%u length=0x%lx ", page, len, pcol->nr_pages, pcol->length); /* split the request, and start again with current page */ |
7aebf4106 exofs: Cleaup rea... |
444 |
ret = read_exec(pcol); |
beaec07ba exofs: address_sp... |
445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 |
if (unlikely(ret)) goto fail; goto try_again; } return 0; fail: /* SetPageError(page); ??? */ unlock_page(page); return ret; } static int exofs_readpages(struct file *file, struct address_space *mapping, struct list_head *pages, unsigned nr_pages) { struct page_collect pcol; int ret; _pcol_init(&pcol, nr_pages, mapping->host); ret = read_cache_pages(mapping, pages, readpage_strip, &pcol); if (ret) { EXOFS_ERR("read_cache_pages => %d ", ret); return ret; } |
b916c5cd4 ore: Only IO one ... |
473 474 475 |
ret = read_exec(&pcol); if (unlikely(ret)) return ret; |
7aebf4106 exofs: Cleaup rea... |
476 |
return read_exec(&pcol); |
beaec07ba exofs: address_sp... |
477 |
} |
7aebf4106 exofs: Cleaup rea... |
478 |
static int _readpage(struct page *page, bool read_4_write) |
beaec07ba exofs: address_sp... |
479 480 481 482 483 |
{ struct page_collect pcol; int ret; _pcol_init(&pcol, 1, page->mapping->host); |
7aebf4106 exofs: Cleaup rea... |
484 |
pcol.read_4_write = read_4_write; |
beaec07ba exofs: address_sp... |
485 486 487 488 489 490 |
ret = readpage_strip(&pcol, page); if (ret) { EXOFS_ERR("_readpage => %d ", ret); return ret; } |
7aebf4106 exofs: Cleaup rea... |
491 |
return read_exec(&pcol); |
beaec07ba exofs: address_sp... |
492 493 494 495 496 497 498 499 500 |
} /* * We don't need the file */ static int exofs_readpage(struct file *file, struct page *page) { return _readpage(page, false); } |
06886a5a3 exofs: Move all o... |
501 |
/* Callback for osd_write. All writes are asynchronous */ |
8ff660ab8 exofs: Rename rai... |
502 |
static void writepages_done(struct ore_io_state *ios, void *p) |
beaec07ba exofs: address_sp... |
503 504 |
{ struct page_collect *pcol = p; |
beaec07ba exofs: address_sp... |
505 |
int i; |
beaec07ba exofs: address_sp... |
506 507 |
u64 good_bytes; u64 length = 0; |
4b46c9f5c ore/exofs: Change... |
508 |
int ret = ore_check_io(ios, NULL); |
beaec07ba exofs: address_sp... |
509 |
|
beaec07ba exofs: address_sp... |
510 |
atomic_dec(&pcol->sbi->s_curr_pending); |
154a9300c exofs: Support fo... |
511 |
if (likely(!ret)) { |
beaec07ba exofs: address_sp... |
512 |
good_bytes = pcol->length; |
154a9300c exofs: Support fo... |
513 514 |
ret = PAGE_WAS_NOT_IN_IO; } else { |
4b46c9f5c ore/exofs: Change... |
515 |
good_bytes = 0; |
154a9300c exofs: Support fo... |
516 |
} |
beaec07ba exofs: address_sp... |
517 |
|
34ce4e7c2 exofs: debug prin... |
518 |
EXOFS_DBGMSG2("writepages_done(0x%lx) good_bytes=0x%llx" |
beaec07ba exofs: address_sp... |
519 520 521 522 |
" length=0x%lx nr_pages=%u ", pcol->inode->i_ino, _LLU(good_bytes), pcol->length, pcol->nr_pages); |
86093aaff exofs: convert io... |
523 524 |
for (i = 0; i < pcol->nr_pages; i++) { struct page *page = pcol->pages[i]; |
beaec07ba exofs: address_sp... |
525 526 527 528 529 530 531 532 533 534 535 536 537 |
struct inode *inode = page->mapping->host; int page_stat; if (inode != pcol->inode) continue; /* osd might add more pages to a bio */ if (likely(length < good_bytes)) page_stat = 0; else page_stat = ret; update_write_page(page, page_stat); unlock_page(page); |
fe33cc1ee exofs: dbg-print ... |
538 539 |
EXOFS_DBGMSG2(" writepages_done(0x%lx, 0x%lx) status=%d ", |
beaec07ba exofs: address_sp... |
540 |
inode->i_ino, page->index, page_stat); |
86093aaff exofs: convert io... |
541 |
length += PAGE_SIZE; |
beaec07ba exofs: address_sp... |
542 543 544 545 |
} pcol_free(pcol); kfree(pcol); |
34ce4e7c2 exofs: debug prin... |
546 547 |
EXOFS_DBGMSG2("writepages_done END "); |
beaec07ba exofs: address_sp... |
548 |
} |
dd2966199 exofs: Support fo... |
549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 |
static struct page *__r4w_get_page(void *priv, u64 offset, bool *uptodate) { struct page_collect *pcol = priv; pgoff_t index = offset / PAGE_SIZE; if (!pcol->that_locked_page || (pcol->that_locked_page->index != index)) { struct page *page = find_get_page(pcol->inode->i_mapping, index); if (!page) { page = find_or_create_page(pcol->inode->i_mapping, index, GFP_NOFS); if (unlikely(!page)) { EXOFS_DBGMSG("grab_cache_page Failed " "index=0x%llx ", _LLU(index)); return NULL; } unlock_page(page); } if (PageDirty(page) || PageWriteback(page)) *uptodate = true; else *uptodate = PageUptodate(page); EXOFS_DBGMSG("index=0x%lx uptodate=%d ", index, *uptodate); return page; } else { EXOFS_DBGMSG("YES that_locked_page index=0x%lx ", pcol->that_locked_page->index); *uptodate = true; return pcol->that_locked_page; } } static void __r4w_put_page(void *priv, struct page *page) { struct page_collect *pcol = priv; if (pcol->that_locked_page != page) { EXOFS_DBGMSG("index=0x%lx ", page->index); page_cache_release(page); return; } EXOFS_DBGMSG("that_locked_page index=0x%lx ", page->index); } static const struct _ore_r4w_op _r4w_op = { .get_page = &__r4w_get_page, .put_page = &__r4w_put_page, }; |
beaec07ba exofs: address_sp... |
603 604 605 |
static int write_exec(struct page_collect *pcol) { struct exofs_i_info *oi = exofs_i(pcol->inode); |
8ff660ab8 exofs: Rename rai... |
606 |
struct ore_io_state *ios; |
beaec07ba exofs: address_sp... |
607 |
struct page_collect *pcol_copy = NULL; |
beaec07ba exofs: address_sp... |
608 |
int ret; |
86093aaff exofs: convert io... |
609 |
if (!pcol->pages) |
beaec07ba exofs: address_sp... |
610 |
return 0; |
e1042ba09 exofs: Add offset... |
611 |
BUG_ON(pcol->ios); |
5bf696dad exofs: Rename str... |
612 |
ret = ore_get_rw_state(&pcol->sbi->layout, &oi->oc, false, |
e1042ba09 exofs: Add offset... |
613 614 |
pcol->pg_first << PAGE_CACHE_SHIFT, pcol->length, &pcol->ios); |
e1042ba09 exofs: Add offset... |
615 616 |
if (unlikely(ret)) goto err; |
beaec07ba exofs: address_sp... |
617 618 |
pcol_copy = kmalloc(sizeof(*pcol_copy), GFP_KERNEL); if (!pcol_copy) { |
571f7f46b fs/exofs: typo fi... |
619 620 |
EXOFS_ERR("write_exec: Failed to kmalloc(pcol) "); |
beaec07ba exofs: address_sp... |
621 622 623 624 625 |
ret = -ENOMEM; goto err; } *pcol_copy = *pcol; |
e1042ba09 exofs: Add offset... |
626 |
ios = pcol->ios; |
86093aaff exofs: convert io... |
627 |
ios->pages = pcol_copy->pages; |
06886a5a3 exofs: Move all o... |
628 |
ios->done = writepages_done; |
dd2966199 exofs: Support fo... |
629 |
ios->r4w = &_r4w_op; |
06886a5a3 exofs: Move all o... |
630 |
ios->private = pcol_copy; |
b916c5cd4 ore: Only IO one ... |
631 632 633 634 635 636 637 638 639 640 |
/* pages ownership was passed to pcol_copy */ _pcol_reset(pcol); ret = _maybe_not_all_in_one_io(ios, pcol_copy, pcol); if (unlikely(ret)) goto err; EXOFS_DBGMSG2("write_exec(0x%lx) offset=0x%llx length=0x%llx ", pcol->inode->i_ino, _LLU(ios->offset), _LLU(ios->length)); |
8ff660ab8 exofs: Rename rai... |
641 |
ret = ore_write(ios); |
beaec07ba exofs: address_sp... |
642 |
if (unlikely(ret)) { |
8ff660ab8 exofs: Rename rai... |
643 644 |
EXOFS_ERR("write_exec: ore_write() Failed "); |
beaec07ba exofs: address_sp... |
645 646 647 648 |
goto err; } atomic_inc(&pcol->sbi->s_curr_pending); |
beaec07ba exofs: address_sp... |
649 650 651 652 |
return 0; err: _unlock_pcol_pages(pcol, ret, WRITE); |
06886a5a3 exofs: Move all o... |
653 |
pcol_free(pcol); |
beaec07ba exofs: address_sp... |
654 |
kfree(pcol_copy); |
06886a5a3 exofs: Move all o... |
655 |
|
beaec07ba exofs: address_sp... |
656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 |
return ret; } /* writepage_strip is called either directly from writepage() or by the VFS from * within write_cache_pages(), to add one more page to be written to storage. * It will try to collect as many contiguous pages as possible. If a * discontinuity is encountered or it runs out of resources it will submit the * previous segment and will start a new collection. * Eventually caller must submit the last segment if present. */ static int writepage_strip(struct page *page, struct writeback_control *wbc_unused, void *data) { struct page_collect *pcol = data; struct inode *inode = pcol->inode; struct exofs_i_info *oi = exofs_i(inode); loff_t i_size = i_size_read(inode); pgoff_t end_index = i_size >> PAGE_CACHE_SHIFT; size_t len; int ret; BUG_ON(!PageLocked(page)); ret = wait_obj_created(oi); if (unlikely(ret)) goto fail; if (page->index < end_index) /* in this case, the page is within the limits of the file */ len = PAGE_CACHE_SIZE; else { len = i_size & ~PAGE_CACHE_MASK; if (page->index > end_index || !len) { /* in this case, the page is outside the limits * (truncate in progress) */ ret = write_exec(pcol); if (unlikely(ret)) goto fail; if (PageError(page)) ClearPageError(page); unlock_page(page); |
06886a5a3 exofs: Move all o... |
699 700 701 702 |
EXOFS_DBGMSG("writepage_strip(0x%lx, 0x%lx) " "outside the limits ", inode->i_ino, page->index); |
beaec07ba exofs: address_sp... |
703 704 705 706 707 708 709 710 711 712 713 714 715 716 |
return 0; } } try_again: if (unlikely(pcol->pg_first == -1)) { pcol->pg_first = page->index; } else if (unlikely((pcol->pg_first + pcol->nr_pages) != page->index)) { /* Discontinuity detected, split the request */ ret = write_exec(pcol); if (unlikely(ret)) goto fail; |
06886a5a3 exofs: Move all o... |
717 718 719 720 |
EXOFS_DBGMSG("writepage_strip(0x%lx, 0x%lx) Discontinuity ", inode->i_ino, page->index); |
beaec07ba exofs: address_sp... |
721 722 |
goto try_again; } |
86093aaff exofs: convert io... |
723 |
if (!pcol->pages) { |
beaec07ba exofs: address_sp... |
724 725 726 727 |
ret = pcol_try_alloc(pcol); if (unlikely(ret)) goto fail; } |
fe33cc1ee exofs: dbg-print ... |
728 729 |
EXOFS_DBGMSG2(" writepage_strip(0x%lx, 0x%lx) len=0x%zx ", |
beaec07ba exofs: address_sp... |
730 731 732 733 |
inode->i_ino, page->index, len); ret = pcol_add_page(pcol, page, len); if (unlikely(ret)) { |
34ce4e7c2 exofs: debug prin... |
734 |
EXOFS_DBGMSG2("Failed pcol_add_page " |
beaec07ba exofs: address_sp... |
735 736 737 738 739 740 741 |
"nr_pages=%u total_length=0x%lx ", pcol->nr_pages, pcol->length); /* split the request, next loop will start again */ ret = write_exec(pcol); if (unlikely(ret)) { |
571f7f46b fs/exofs: typo fi... |
742 |
EXOFS_DBGMSG("write_exec failed => %d", ret); |
beaec07ba exofs: address_sp... |
743 744 745 746 747 748 749 750 751 752 753 754 |
goto fail; } goto try_again; } BUG_ON(PageWriteback(page)); set_page_writeback(page); return 0; fail: |
06886a5a3 exofs: Move all o... |
755 756 757 |
EXOFS_DBGMSG("Error: writepage_strip(0x%lx, 0x%lx)=>%d ", inode->i_ino, page->index, ret); |
beaec07ba exofs: address_sp... |
758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 |
set_bit(AS_EIO, &page->mapping->flags); unlock_page(page); return ret; } static int exofs_writepages(struct address_space *mapping, struct writeback_control *wbc) { struct page_collect pcol; long start, end, expected_pages; int ret; start = wbc->range_start >> PAGE_CACHE_SHIFT; end = (wbc->range_end == LLONG_MAX) ? start + mapping->nrpages : wbc->range_end >> PAGE_CACHE_SHIFT; if (start || end) |
06886a5a3 exofs: Move all o... |
776 |
expected_pages = end - start + 1; |
beaec07ba exofs: address_sp... |
777 778 |
else expected_pages = mapping->nrpages; |
06886a5a3 exofs: Move all o... |
779 780 |
if (expected_pages < 32L) expected_pages = 32L; |
34ce4e7c2 exofs: debug prin... |
781 |
EXOFS_DBGMSG2("inode(0x%lx) wbc->start=0x%llx wbc->end=0x%llx " |
06886a5a3 exofs: Move all o... |
782 783 |
"nrpages=%lu start=0x%lx end=0x%lx expected_pages=%ld ", |
beaec07ba exofs: address_sp... |
784 |
mapping->host->i_ino, wbc->range_start, wbc->range_end, |
06886a5a3 exofs: Move all o... |
785 |
mapping->nrpages, start, end, expected_pages); |
beaec07ba exofs: address_sp... |
786 787 788 789 |
_pcol_init(&pcol, expected_pages, mapping->host); ret = write_cache_pages(mapping, wbc, writepage_strip, &pcol); |
b916c5cd4 ore: Only IO one ... |
790 |
if (unlikely(ret)) { |
beaec07ba exofs: address_sp... |
791 792 793 794 |
EXOFS_ERR("write_cache_pages => %d ", ret); return ret; } |
b916c5cd4 ore: Only IO one ... |
795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 |
ret = write_exec(&pcol); if (unlikely(ret)) return ret; if (wbc->sync_mode == WB_SYNC_ALL) { return write_exec(&pcol); /* pump the last reminder */ } else if (pcol.nr_pages) { /* not SYNC let the reminder join the next writeout */ unsigned i; for (i = 0; i < pcol.nr_pages; i++) { struct page *page = pcol.pages[i]; end_page_writeback(page); set_page_dirty(page); unlock_page(page); } } return 0; |
beaec07ba exofs: address_sp... |
814 |
} |
dd2966199 exofs: Support fo... |
815 |
/* |
beaec07ba exofs: address_sp... |
816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 |
static int exofs_writepage(struct page *page, struct writeback_control *wbc) { struct page_collect pcol; int ret; _pcol_init(&pcol, 1, page->mapping->host); ret = writepage_strip(page, NULL, &pcol); if (ret) { EXOFS_ERR("exofs_writepage => %d ", ret); return ret; } return write_exec(&pcol); } |
dd2966199 exofs: Support fo... |
832 |
*/ |
2f246fd0f exofs: New trunca... |
833 834 835 836 837 838 |
/* i_mutex held using inode->i_size directly */ static void _write_failed(struct inode *inode, loff_t to) { if (to > inode->i_size) truncate_pagecache(inode, to, inode->i_size); } |
beaec07ba exofs: address_sp... |
839 840 841 842 843 844 845 846 847 848 849 850 |
int exofs_write_begin(struct file *file, struct address_space *mapping, loff_t pos, unsigned len, unsigned flags, struct page **pagep, void **fsdata) { int ret = 0; struct page *page; page = *pagep; if (page == NULL) { ret = simple_write_begin(file, mapping, pos, len, flags, pagep, fsdata); if (ret) { |
571f7f46b fs/exofs: typo fi... |
851 852 |
EXOFS_DBGMSG("simple_write_begin failed "); |
2f246fd0f exofs: New trunca... |
853 |
goto out; |
beaec07ba exofs: address_sp... |
854 855 856 857 858 859 860 |
} page = *pagep; } /* read modify write */ if (!PageUptodate(page) && (len != PAGE_CACHE_SIZE)) { |
a8f1418f9 exofs: Optimize r... |
861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 |
loff_t i_size = i_size_read(mapping->host); pgoff_t end_index = i_size >> PAGE_CACHE_SHIFT; size_t rlen; if (page->index < end_index) rlen = PAGE_CACHE_SIZE; else if (page->index == end_index) rlen = i_size & ~PAGE_CACHE_MASK; else rlen = 0; if (!rlen) { clear_highpage(page); SetPageUptodate(page); goto out; } |
beaec07ba exofs: address_sp... |
877 878 879 880 |
ret = _readpage(page, true); if (ret) { /*SetPageError was done by _readpage. Is it ok?*/ unlock_page(page); |
a8f1418f9 exofs: Optimize r... |
881 882 |
EXOFS_DBGMSG("__readpage failed "); |
beaec07ba exofs: address_sp... |
883 884 |
} } |
2f246fd0f exofs: New trunca... |
885 886 887 |
out: if (unlikely(ret)) _write_failed(mapping->host, pos + len); |
beaec07ba exofs: address_sp... |
888 889 890 891 892 893 894 895 896 897 898 899 900 901 |
return ret; } static int exofs_write_begin_export(struct file *file, struct address_space *mapping, loff_t pos, unsigned len, unsigned flags, struct page **pagep, void **fsdata) { *pagep = NULL; return exofs_write_begin(file, mapping, pos, len, flags, pagep, fsdata); } |
efd124b99 exofs: simple_wri... |
902 903 904 905 906 907 908 909 910 911 |
static int exofs_write_end(struct file *file, struct address_space *mapping, loff_t pos, unsigned len, unsigned copied, struct page *page, void *fsdata) { struct inode *inode = mapping->host; /* According to comment in simple_write_end i_mutex is held */ loff_t i_size = inode->i_size; int ret; ret = simple_write_end(file, mapping,pos, len, copied, page, fsdata); |
2f246fd0f exofs: New trunca... |
912 913 914 915 |
if (unlikely(ret)) _write_failed(inode, pos + len); /* TODO: once simple_write_end marks inode dirty remove */ |
efd124b99 exofs: simple_wri... |
916 917 918 919 |
if (i_size != inode->i_size) mark_inode_dirty(inode); return ret; } |
200b07004 exofs: Add defaul... |
920 921 922 923 924 |
static int exofs_releasepage(struct page *page, gfp_t gfp) { EXOFS_DBGMSG("page 0x%lx ", page->index); WARN_ON(1); |
85dc7878c exofs: Remove sup... |
925 |
return 0; |
200b07004 exofs: Add defaul... |
926 927 928 929 |
} static void exofs_invalidatepage(struct page *page, unsigned long offset) { |
85dc7878c exofs: Remove sup... |
930 931 |
EXOFS_DBGMSG("page 0x%lx offset 0x%lx ", page->index, offset); |
200b07004 exofs: Add defaul... |
932 |
WARN_ON(1); |
200b07004 exofs: Add defaul... |
933 |
} |
beaec07ba exofs: address_sp... |
934 935 936 |
const struct address_space_operations exofs_aops = { .readpage = exofs_readpage, .readpages = exofs_readpages, |
dd2966199 exofs: Support fo... |
937 |
.writepage = NULL, |
beaec07ba exofs: address_sp... |
938 939 |
.writepages = exofs_writepages, .write_begin = exofs_write_begin_export, |
efd124b99 exofs: simple_wri... |
940 |
.write_end = exofs_write_end, |
200b07004 exofs: Add defaul... |
941 942 943 944 945 946 947 948 949 |
.releasepage = exofs_releasepage, .set_page_dirty = __set_page_dirty_nobuffers, .invalidatepage = exofs_invalidatepage, /* Not implemented Yet */ .bmap = NULL, /* TODO: use osd's OSD_ACT_READ_MAP */ .direct_IO = NULL, /* TODO: Should be trivial to do */ /* With these NULL has special meaning or default is not exported */ |
200b07004 exofs: Add defaul... |
950 951 952 953 954 |
.get_xip_mem = NULL, .migratepage = NULL, .launder_page = NULL, .is_partially_uptodate = NULL, .error_remove_page = NULL, |
beaec07ba exofs: address_sp... |
955 |
}; |
e80627191 exofs: file and f... |
956 957 958 959 960 961 962 963 964 965 966 967 968 |
/****************************************************************************** * INODE OPERATIONS *****************************************************************************/ /* * Test whether an inode is a fast symlink. */ static inline int exofs_inode_is_fast_symlink(struct inode *inode) { struct exofs_i_info *oi = exofs_i(inode); return S_ISLNK(inode->i_mode) && (oi->i_data[0] != 0); } |
2f246fd0f exofs: New trunca... |
969 |
static int _do_truncate(struct inode *inode, loff_t newsize) |
06886a5a3 exofs: Move all o... |
970 971 |
{ struct exofs_i_info *oi = exofs_i(inode); |
9e9db4564 exofs: ios: Move ... |
972 |
struct exofs_sb_info *sbi = inode->i_sb->s_fs_info; |
06886a5a3 exofs: Move all o... |
973 974 975 |
int ret; inode->i_mtime = inode->i_ctime = CURRENT_TIME; |
5bf696dad exofs: Rename str... |
976 |
ret = ore_truncate(&sbi->layout, &oi->oc, (u64)newsize); |
2f246fd0f exofs: New trunca... |
977 978 |
if (likely(!ret)) truncate_setsize(inode, newsize); |
06886a5a3 exofs: Move all o... |
979 |
|
2f246fd0f exofs: New trunca... |
980 981 982 |
EXOFS_DBGMSG("(0x%lx) size=0x%llx ret=>%d ", inode->i_ino, newsize, ret); |
06886a5a3 exofs: Move all o... |
983 984 |
return ret; } |
e80627191 exofs: file and f... |
985 |
/* |
2f246fd0f exofs: New trunca... |
986 987 |
* Set inode attributes - update size attribute on OSD if needed, * otherwise just call generic functions. |
e80627191 exofs: file and f... |
988 989 990 991 992 |
*/ int exofs_setattr(struct dentry *dentry, struct iattr *iattr) { struct inode *inode = dentry->d_inode; int error; |
2f246fd0f exofs: New trunca... |
993 994 995 996 997 998 |
/* if we are about to modify an object, and it hasn't been * created yet, wait */ error = wait_obj_created(exofs_i(inode)); if (unlikely(error)) return error; |
e80627191 exofs: file and f... |
999 |
error = inode_change_ok(inode, iattr); |
2f246fd0f exofs: New trunca... |
1000 |
if (unlikely(error)) |
e80627191 exofs: file and f... |
1001 |
return error; |
1025774ce remove inode_setattr |
1002 1003 |
if ((iattr->ia_valid & ATTR_SIZE) && iattr->ia_size != i_size_read(inode)) { |
2f246fd0f exofs: New trunca... |
1004 1005 |
error = _do_truncate(inode, iattr->ia_size); if (unlikely(error)) |
1025774ce remove inode_setattr |
1006 1007 1008 1009 1010 1011 |
return error; } setattr_copy(inode, iattr); mark_inode_dirty(inode); return 0; |
e80627191 exofs: file and f... |
1012 |
} |
e6af00f1d exofs: dir_inode ... |
1013 |
|
d9c740d22 exofs: Define on-... |
1014 1015 1016 1017 1018 1019 1020 1021 |
static const struct osd_attr g_attr_inode_file_layout = ATTR_DEF( EXOFS_APAGE_FS_DATA, EXOFS_ATTR_INODE_FILE_LAYOUT, 0); static const struct osd_attr g_attr_inode_dir_layout = ATTR_DEF( EXOFS_APAGE_FS_DATA, EXOFS_ATTR_INODE_DIR_LAYOUT, 0); |
e6af00f1d exofs: dir_inode ... |
1022 |
/* |
5d952b839 exofs: RAID0 support |
1023 1024 |
* Read the Linux inode info from the OSD, and return it as is. In exofs the * inode info is in an application specific page/attribute of the osd-object. |
e6af00f1d exofs: dir_inode ... |
1025 1026 |
*/ static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi, |
5d952b839 exofs: RAID0 support |
1027 |
struct exofs_fcb *inode) |
e6af00f1d exofs: dir_inode ... |
1028 1029 |
{ struct exofs_sb_info *sbi = sb->s_fs_info; |
d9c740d22 exofs: Define on-... |
1030 1031 1032 1033 |
struct osd_attr attrs[] = { [0] = g_attr_inode_data, [1] = g_attr_inode_file_layout, [2] = g_attr_inode_dir_layout, |
d9c740d22 exofs: Define on-... |
1034 |
}; |
8ff660ab8 exofs: Rename rai... |
1035 |
struct ore_io_state *ios; |
d9c740d22 exofs: Define on-... |
1036 |
struct exofs_on_disk_inode_layout *layout; |
e6af00f1d exofs: dir_inode ... |
1037 |
int ret; |
5bf696dad exofs: Rename str... |
1038 |
ret = ore_get_io_state(&sbi->layout, &oi->oc, &ios); |
06886a5a3 exofs: Move all o... |
1039 |
if (unlikely(ret)) { |
8ff660ab8 exofs: Rename rai... |
1040 1041 |
EXOFS_ERR("%s: ore_get_io_state failed. ", __func__); |
06886a5a3 exofs: Move all o... |
1042 |
return ret; |
e6af00f1d exofs: dir_inode ... |
1043 |
} |
e6af00f1d exofs: dir_inode ... |
1044 |
|
5bf696dad exofs: Rename str... |
1045 1046 |
attrs[1].len = exofs_on_disk_inode_layout_size(sbi->oc.numdevs); attrs[2].len = exofs_on_disk_inode_layout_size(sbi->oc.numdevs); |
d9c740d22 exofs: Define on-... |
1047 |
|
06886a5a3 exofs: Move all o... |
1048 1049 |
ios->in_attr = attrs; ios->in_attr_len = ARRAY_SIZE(attrs); |
e6af00f1d exofs: dir_inode ... |
1050 |
|
8ff660ab8 exofs: Rename rai... |
1051 |
ret = ore_read(ios); |
96391e2ba exofs: Error reco... |
1052 1053 1054 |
if (unlikely(ret)) { EXOFS_ERR("object(0x%llx) corrupted, return empty file=>%d ", |
9e9db4564 exofs: ios: Move ... |
1055 |
_LLU(oi->one_comp.obj.id), ret); |
96391e2ba exofs: Error reco... |
1056 1057 1058 1059 1060 1061 1062 |
memset(inode, 0, sizeof(*inode)); inode->i_mode = 0040000 | (0777 & ~022); /* If object is lost on target we might as well enable it's * delete. */ if ((ret == -ENOENT) || (ret == -EINVAL)) ret = 0; |
e6af00f1d exofs: dir_inode ... |
1063 |
goto out; |
96391e2ba exofs: Error reco... |
1064 |
} |
e6af00f1d exofs: dir_inode ... |
1065 |
|
06886a5a3 exofs: Move all o... |
1066 |
ret = extract_attr_from_ios(ios, &attrs[0]); |
e6af00f1d exofs: dir_inode ... |
1067 |
if (ret) { |
06886a5a3 exofs: Move all o... |
1068 1069 |
EXOFS_ERR("%s: extract_attr of inode_data failed ", __func__); |
e6af00f1d exofs: dir_inode ... |
1070 1071 |
goto out; } |
06886a5a3 exofs: Move all o... |
1072 1073 |
WARN_ON(attrs[0].len != EXOFS_INO_ATTR_SIZE); memcpy(inode, attrs[0].val_ptr, EXOFS_INO_ATTR_SIZE); |
e6af00f1d exofs: dir_inode ... |
1074 |
|
06886a5a3 exofs: Move all o... |
1075 |
ret = extract_attr_from_ios(ios, &attrs[1]); |
e6af00f1d exofs: dir_inode ... |
1076 |
if (ret) { |
d9c740d22 exofs: Define on-... |
1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 |
EXOFS_ERR("%s: extract_attr of inode_data failed ", __func__); goto out; } if (attrs[1].len) { layout = attrs[1].val_ptr; if (layout->gen_func != cpu_to_le16(LAYOUT_MOVING_WINDOW)) { EXOFS_ERR("%s: unsupported files layout %d ", __func__, layout->gen_func); ret = -ENOTSUPP; goto out; } } ret = extract_attr_from_ios(ios, &attrs[2]); if (ret) { EXOFS_ERR("%s: extract_attr of inode_data failed ", __func__); goto out; } if (attrs[2].len) { layout = attrs[2].val_ptr; if (layout->gen_func != cpu_to_le16(LAYOUT_MOVING_WINDOW)) { EXOFS_ERR("%s: unsupported meta-data layout %d ", __func__, layout->gen_func); ret = -ENOTSUPP; goto out; } } |
e6af00f1d exofs: dir_inode ... |
1108 |
out: |
8ff660ab8 exofs: Rename rai... |
1109 |
ore_put_io_state(ios); |
e6af00f1d exofs: dir_inode ... |
1110 1111 |
return ret; } |
9cfdc7aa9 exofs: refactor e... |
1112 1113 1114 1115 1116 |
static void __oi_init(struct exofs_i_info *oi) { init_waitqueue_head(&oi->i_wq); oi->i_flags = 0; } |
e6af00f1d exofs: dir_inode ... |
1117 1118 1119 1120 1121 1122 1123 1124 |
/* * Fill in an inode read from the OSD and set it up for use */ struct inode *exofs_iget(struct super_block *sb, unsigned long ino) { struct exofs_i_info *oi; struct exofs_fcb fcb; struct inode *inode; |
e6af00f1d exofs: dir_inode ... |
1125 1126 1127 1128 1129 1130 1131 1132 |
int ret; inode = iget_locked(sb, ino); if (!inode) return ERR_PTR(-ENOMEM); if (!(inode->i_state & I_NEW)) return inode; oi = exofs_i(inode); |
9cfdc7aa9 exofs: refactor e... |
1133 |
__oi_init(oi); |
5bf696dad exofs: Rename str... |
1134 |
exofs_init_comps(&oi->oc, &oi->one_comp, sb->s_fs_info, |
9e9db4564 exofs: ios: Move ... |
1135 |
exofs_oi_objno(oi)); |
e6af00f1d exofs: dir_inode ... |
1136 1137 |
/* read the inode from the osd */ |
5d952b839 exofs: RAID0 support |
1138 |
ret = exofs_get_inode(sb, oi, &fcb); |
e6af00f1d exofs: dir_inode ... |
1139 1140 |
if (ret) goto bad_inode; |
e6af00f1d exofs: dir_inode ... |
1141 1142 1143 1144 1145 1146 |
set_obj_created(oi); /* copy stuff from on-disk struct to in-memory struct */ inode->i_mode = le16_to_cpu(fcb.i_mode); inode->i_uid = le32_to_cpu(fcb.i_uid); inode->i_gid = le32_to_cpu(fcb.i_gid); |
bfe868486 filesystems: add ... |
1147 |
set_nlink(inode, le16_to_cpu(fcb.i_links_count)); |
e6af00f1d exofs: dir_inode ... |
1148 1149 1150 1151 1152 1153 1154 1155 1156 |
inode->i_ctime.tv_sec = (signed)le32_to_cpu(fcb.i_ctime); inode->i_atime.tv_sec = (signed)le32_to_cpu(fcb.i_atime); inode->i_mtime.tv_sec = (signed)le32_to_cpu(fcb.i_mtime); inode->i_ctime.tv_nsec = inode->i_atime.tv_nsec = inode->i_mtime.tv_nsec = 0; oi->i_commit_size = le64_to_cpu(fcb.i_size); i_size_write(inode, oi->i_commit_size); inode->i_blkbits = EXOFS_BLKSHIFT; inode->i_generation = le32_to_cpu(fcb.i_generation); |
e6af00f1d exofs: dir_inode ... |
1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 |
oi->i_dir_start_lookup = 0; if ((inode->i_nlink == 0) && (inode->i_mode == 0)) { ret = -ESTALE; goto bad_inode; } if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode)) { if (fcb.i_data[0]) inode->i_rdev = old_decode_dev(le32_to_cpu(fcb.i_data[0])); else inode->i_rdev = new_decode_dev(le32_to_cpu(fcb.i_data[1])); } else { memcpy(oi->i_data, fcb.i_data, sizeof(fcb.i_data)); } |
66cd6cad4 exofs: Override r... |
1174 |
inode->i_mapping->backing_dev_info = sb->s_bdi; |
e6af00f1d exofs: dir_inode ... |
1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 |
if (S_ISREG(inode->i_mode)) { inode->i_op = &exofs_file_inode_operations; inode->i_fop = &exofs_file_operations; inode->i_mapping->a_ops = &exofs_aops; } else if (S_ISDIR(inode->i_mode)) { inode->i_op = &exofs_dir_inode_operations; inode->i_fop = &exofs_dir_operations; inode->i_mapping->a_ops = &exofs_aops; } else if (S_ISLNK(inode->i_mode)) { if (exofs_inode_is_fast_symlink(inode)) inode->i_op = &exofs_fast_symlink_inode_operations; else { inode->i_op = &exofs_symlink_inode_operations; inode->i_mapping->a_ops = &exofs_aops; } } else { inode->i_op = &exofs_special_inode_operations; if (fcb.i_data[0]) init_special_inode(inode, inode->i_mode, old_decode_dev(le32_to_cpu(fcb.i_data[0]))); else init_special_inode(inode, inode->i_mode, new_decode_dev(le32_to_cpu(fcb.i_data[1]))); } unlock_new_inode(inode); return inode; bad_inode: iget_failed(inode); return ERR_PTR(ret); } int __exofs_wait_obj_created(struct exofs_i_info *oi) { if (!obj_created(oi)) { |
fe2fd9ed5 exofs: Remove ino... |
1211 1212 |
EXOFS_DBGMSG("!obj_created "); |
e6af00f1d exofs: dir_inode ... |
1213 1214 |
BUG_ON(!obj_2bcreated(oi)); wait_event(oi->i_wq, obj_created(oi)); |
fe2fd9ed5 exofs: Remove ino... |
1215 1216 |
EXOFS_DBGMSG("wait_event done "); |
e6af00f1d exofs: dir_inode ... |
1217 1218 1219 |
} return unlikely(is_bad_inode(&oi->vfs_inode)) ? -EIO : 0; } |
1cea312ad exofs: Write sbi-... |
1220 |
|
e6af00f1d exofs: dir_inode ... |
1221 1222 1223 1224 1225 |
/* * Callback function from exofs_new_inode(). The important thing is that we * set the obj_created flag so that other methods know that the object exists on * the OSD. */ |
8ff660ab8 exofs: Rename rai... |
1226 |
static void create_done(struct ore_io_state *ios, void *p) |
e6af00f1d exofs: dir_inode ... |
1227 1228 1229 1230 1231 |
{ struct inode *inode = p; struct exofs_i_info *oi = exofs_i(inode); struct exofs_sb_info *sbi = inode->i_sb->s_fs_info; int ret; |
8ff660ab8 exofs: Rename rai... |
1232 1233 |
ret = ore_check_io(ios, NULL); ore_put_io_state(ios); |
06886a5a3 exofs: Move all o... |
1234 |
|
e6af00f1d exofs: dir_inode ... |
1235 1236 1237 |
atomic_dec(&sbi->s_curr_pending); if (unlikely(ret)) { |
571f7f46b fs/exofs: typo fi... |
1238 |
EXOFS_ERR("object=0x%llx creation failed in pid=0x%llx", |
9e9db4564 exofs: ios: Move ... |
1239 1240 |
_LLU(exofs_oi_objno(oi)), _LLU(oi->one_comp.obj.partition)); |
06886a5a3 exofs: Move all o... |
1241 1242 1243 1244 1245 1246 1247 1248 1249 |
/*TODO: When FS is corrupted creation can fail, object already * exist. Get rid of this asynchronous creation, if exist * increment the obj counter and try the next object. Until we * succeed. All these dangling objects will be made into lost * files by chkfs.exofs */ } set_obj_created(oi); |
e6af00f1d exofs: dir_inode ... |
1250 |
|
e6af00f1d exofs: dir_inode ... |
1251 1252 1253 1254 1255 1256 1257 1258 |
wake_up(&oi->i_wq); } /* * Set up a new inode and create an object for it on the OSD */ struct inode *exofs_new_inode(struct inode *dir, int mode) { |
9e9db4564 exofs: ios: Move ... |
1259 1260 |
struct super_block *sb = dir->i_sb; struct exofs_sb_info *sbi = sb->s_fs_info; |
e6af00f1d exofs: dir_inode ... |
1261 1262 |
struct inode *inode; struct exofs_i_info *oi; |
8ff660ab8 exofs: Rename rai... |
1263 |
struct ore_io_state *ios; |
e6af00f1d exofs: dir_inode ... |
1264 |
int ret; |
e6af00f1d exofs: dir_inode ... |
1265 1266 1267 1268 1269 |
inode = new_inode(sb); if (!inode) return ERR_PTR(-ENOMEM); oi = exofs_i(inode); |
9cfdc7aa9 exofs: refactor e... |
1270 |
__oi_init(oi); |
e6af00f1d exofs: dir_inode ... |
1271 |
|
e6af00f1d exofs: dir_inode ... |
1272 |
set_obj_2bcreated(oi); |
66cd6cad4 exofs: Override r... |
1273 |
inode->i_mapping->backing_dev_info = sb->s_bdi; |
e00117f14 exofs: replace in... |
1274 |
inode_init_owner(inode, dir, mode); |
e6af00f1d exofs: dir_inode ... |
1275 1276 1277 1278 1279 1280 1281 1282 |
inode->i_ino = sbi->s_nextid++; inode->i_blkbits = EXOFS_BLKSHIFT; inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME; oi->i_commit_size = inode->i_size = 0; spin_lock(&sbi->s_next_gen_lock); inode->i_generation = sbi->s_next_generation++; spin_unlock(&sbi->s_next_gen_lock); insert_inode_hash(inode); |
5bf696dad exofs: Rename str... |
1283 |
exofs_init_comps(&oi->oc, &oi->one_comp, sb->s_fs_info, |
9e9db4564 exofs: ios: Move ... |
1284 |
exofs_oi_objno(oi)); |
1cea312ad exofs: Write sbi-... |
1285 |
exofs_sbi_write_stats(sbi); /* Make sure new sbi->s_nextid is on disk */ |
e6af00f1d exofs: dir_inode ... |
1286 |
mark_inode_dirty(inode); |
5bf696dad exofs: Rename str... |
1287 |
ret = ore_get_io_state(&sbi->layout, &oi->oc, &ios); |
06886a5a3 exofs: Move all o... |
1288 |
if (unlikely(ret)) { |
8ff660ab8 exofs: Rename rai... |
1289 1290 |
EXOFS_ERR("exofs_new_inode: ore_get_io_state failed "); |
06886a5a3 exofs: Move all o... |
1291 |
return ERR_PTR(ret); |
e6af00f1d exofs: dir_inode ... |
1292 |
} |
06886a5a3 exofs: Move all o... |
1293 1294 |
ios->done = create_done; ios->private = inode; |
9e9db4564 exofs: ios: Move ... |
1295 |
|
8ff660ab8 exofs: Rename rai... |
1296 |
ret = ore_create(ios); |
e6af00f1d exofs: dir_inode ... |
1297 |
if (ret) { |
8ff660ab8 exofs: Rename rai... |
1298 |
ore_put_io_state(ios); |
06886a5a3 exofs: Move all o... |
1299 |
return ERR_PTR(ret); |
e6af00f1d exofs: dir_inode ... |
1300 1301 1302 1303 1304 |
} atomic_inc(&sbi->s_curr_pending); return inode; } |
ba9e5e98c exofs: super_oper... |
1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 |
/* * struct to pass two arguments to update_inode's callback */ struct updatei_args { struct exofs_sb_info *sbi; struct exofs_fcb fcb; }; /* * Callback function from exofs_update_inode(). */ |
8ff660ab8 exofs: Rename rai... |
1317 |
static void updatei_done(struct ore_io_state *ios, void *p) |
ba9e5e98c exofs: super_oper... |
1318 1319 |
{ struct updatei_args *args = p; |
8ff660ab8 exofs: Rename rai... |
1320 |
ore_put_io_state(ios); |
ba9e5e98c exofs: super_oper... |
1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 |
atomic_dec(&args->sbi->s_curr_pending); kfree(args); } /* * Write the inode to the OSD. Just fill up the struct, and set the attribute * synchronously or asynchronously depending on the do_sync flag. */ static int exofs_update_inode(struct inode *inode, int do_sync) { struct exofs_i_info *oi = exofs_i(inode); struct super_block *sb = inode->i_sb; struct exofs_sb_info *sbi = sb->s_fs_info; |
8ff660ab8 exofs: Rename rai... |
1336 |
struct ore_io_state *ios; |
ba9e5e98c exofs: super_oper... |
1337 1338 1339 1340 1341 1342 |
struct osd_attr attr; struct exofs_fcb *fcb; struct updatei_args *args; int ret; args = kzalloc(sizeof(*args), GFP_KERNEL); |
34ce4e7c2 exofs: debug prin... |
1343 |
if (!args) { |
571f7f46b fs/exofs: typo fi... |
1344 1345 |
EXOFS_DBGMSG("Failed kzalloc of args "); |
ba9e5e98c exofs: super_oper... |
1346 |
return -ENOMEM; |
34ce4e7c2 exofs: debug prin... |
1347 |
} |
ba9e5e98c exofs: super_oper... |
1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 |
fcb = &args->fcb; fcb->i_mode = cpu_to_le16(inode->i_mode); fcb->i_uid = cpu_to_le32(inode->i_uid); fcb->i_gid = cpu_to_le32(inode->i_gid); fcb->i_links_count = cpu_to_le16(inode->i_nlink); fcb->i_ctime = cpu_to_le32(inode->i_ctime.tv_sec); fcb->i_atime = cpu_to_le32(inode->i_atime.tv_sec); fcb->i_mtime = cpu_to_le32(inode->i_mtime.tv_sec); oi->i_commit_size = i_size_read(inode); fcb->i_size = cpu_to_le64(oi->i_commit_size); fcb->i_generation = cpu_to_le32(inode->i_generation); if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode)) { if (old_valid_dev(inode->i_rdev)) { fcb->i_data[0] = cpu_to_le32(old_encode_dev(inode->i_rdev)); fcb->i_data[1] = 0; } else { fcb->i_data[0] = 0; fcb->i_data[1] = cpu_to_le32(new_encode_dev(inode->i_rdev)); fcb->i_data[2] = 0; } } else memcpy(fcb->i_data, oi->i_data, sizeof(fcb->i_data)); |
5bf696dad exofs: Rename str... |
1375 |
ret = ore_get_io_state(&sbi->layout, &oi->oc, &ios); |
06886a5a3 exofs: Move all o... |
1376 |
if (unlikely(ret)) { |
8ff660ab8 exofs: Rename rai... |
1377 1378 |
EXOFS_ERR("%s: ore_get_io_state failed. ", __func__); |
ba9e5e98c exofs: super_oper... |
1379 1380 |
goto free_args; } |
ba9e5e98c exofs: super_oper... |
1381 1382 |
attr = g_attr_inode_data; attr.val_ptr = fcb; |
06886a5a3 exofs: Move all o... |
1383 1384 |
ios->out_attr_len = 1; ios->out_attr = &attr; |
ba9e5e98c exofs: super_oper... |
1385 |
|
fe2fd9ed5 exofs: Remove ino... |
1386 |
wait_obj_created(oi); |
ba9e5e98c exofs: super_oper... |
1387 |
|
06886a5a3 exofs: Move all o... |
1388 |
if (!do_sync) { |
ba9e5e98c exofs: super_oper... |
1389 |
args->sbi = sbi; |
06886a5a3 exofs: Move all o... |
1390 1391 1392 |
ios->done = updatei_done; ios->private = args; } |
ba9e5e98c exofs: super_oper... |
1393 |
|
8ff660ab8 exofs: Rename rai... |
1394 |
ret = ore_write(ios); |
06886a5a3 exofs: Move all o... |
1395 |
if (!do_sync && !ret) { |
ba9e5e98c exofs: super_oper... |
1396 1397 1398 |
atomic_inc(&sbi->s_curr_pending); goto out; /* deallocation in updatei_done */ } |
8ff660ab8 exofs: Rename rai... |
1399 |
ore_put_io_state(ios); |
ba9e5e98c exofs: super_oper... |
1400 1401 1402 |
free_args: kfree(args); out: |
34ce4e7c2 exofs: debug prin... |
1403 1404 1405 |
EXOFS_DBGMSG("(0x%lx) do_sync=%d ret=>%d ", inode->i_ino, do_sync, ret); |
ba9e5e98c exofs: super_oper... |
1406 1407 |
return ret; } |
a9185b41a pass writeback_co... |
1408 |
int exofs_write_inode(struct inode *inode, struct writeback_control *wbc) |
ba9e5e98c exofs: super_oper... |
1409 |
{ |
97178b7b6 exofs: simple fsy... |
1410 1411 |
/* FIXME: fix fsync and use wbc->sync_mode == WB_SYNC_ALL */ return exofs_update_inode(inode, 1); |
ba9e5e98c exofs: super_oper... |
1412 1413 1414 1415 1416 1417 |
} /* * Callback function from exofs_delete_inode() - don't have much cleaning up to * do. */ |
8ff660ab8 exofs: Rename rai... |
1418 |
static void delete_done(struct ore_io_state *ios, void *p) |
ba9e5e98c exofs: super_oper... |
1419 |
{ |
06886a5a3 exofs: Move all o... |
1420 |
struct exofs_sb_info *sbi = p; |
8ff660ab8 exofs: Rename rai... |
1421 |
ore_put_io_state(ios); |
06886a5a3 exofs: Move all o... |
1422 |
|
ba9e5e98c exofs: super_oper... |
1423 1424 1425 1426 1427 1428 1429 1430 |
atomic_dec(&sbi->s_curr_pending); } /* * Called when the refcount of an inode reaches zero. We remove the object * from the OSD here. We make sure the object was created before we try and * delete it. */ |
4ec70c9b4 convert exofs to ... |
1431 |
void exofs_evict_inode(struct inode *inode) |
ba9e5e98c exofs: super_oper... |
1432 1433 1434 1435 |
{ struct exofs_i_info *oi = exofs_i(inode); struct super_block *sb = inode->i_sb; struct exofs_sb_info *sbi = sb->s_fs_info; |
8ff660ab8 exofs: Rename rai... |
1436 |
struct ore_io_state *ios; |
ba9e5e98c exofs: super_oper... |
1437 1438 1439 |
int ret; truncate_inode_pages(&inode->i_data, 0); |
2f246fd0f exofs: New trunca... |
1440 |
/* TODO: should do better here */ |
4ec70c9b4 convert exofs to ... |
1441 |
if (inode->i_nlink || is_bad_inode(inode)) |
ba9e5e98c exofs: super_oper... |
1442 |
goto no_delete; |
ba9e5e98c exofs: super_oper... |
1443 |
inode->i_size = 0; |
4ec70c9b4 convert exofs to ... |
1444 |
end_writeback(inode); |
ba9e5e98c exofs: super_oper... |
1445 |
|
fe2fd9ed5 exofs: Remove ino... |
1446 1447 1448 1449 1450 1451 |
/* if we are deleting an obj that hasn't been created yet, wait. * This also makes sure that create_done cannot be called with an * already evicted inode. */ wait_obj_created(oi); /* ignore the error, attempt a remove anyway */ |
2f246fd0f exofs: New trunca... |
1452 1453 |
/* Now Remove the OSD objects */ |
5bf696dad exofs: Rename str... |
1454 |
ret = ore_get_io_state(&sbi->layout, &oi->oc, &ios); |
2f246fd0f exofs: New trunca... |
1455 |
if (unlikely(ret)) { |
8ff660ab8 exofs: Rename rai... |
1456 1457 |
EXOFS_ERR("%s: ore_get_io_state failed ", __func__); |
2f246fd0f exofs: New trunca... |
1458 |
return; |
ba9e5e98c exofs: super_oper... |
1459 |
} |
06886a5a3 exofs: Move all o... |
1460 1461 |
ios->done = delete_done; ios->private = sbi; |
9e9db4564 exofs: ios: Move ... |
1462 |
|
8ff660ab8 exofs: Rename rai... |
1463 |
ret = ore_remove(ios); |
ba9e5e98c exofs: super_oper... |
1464 |
if (ret) { |
8ff660ab8 exofs: Rename rai... |
1465 1466 1467 |
EXOFS_ERR("%s: ore_remove failed ", __func__); ore_put_io_state(ios); |
ba9e5e98c exofs: super_oper... |
1468 1469 1470 1471 1472 1473 1474 |
return; } atomic_inc(&sbi->s_curr_pending); return; no_delete: |
4ec70c9b4 convert exofs to ... |
1475 |
end_writeback(inode); |
ba9e5e98c exofs: super_oper... |
1476 |
} |