| 1 |
// SPDX-License-Identifier: GPL-2.0-or-later |
| 2 |
/* |
| 3 |
* Copyright (C) International Business Machines Corp., 2000-2005 |
| 4 |
* Portions Copyright (C) Christoph Hellwig, 2001-2002 |
| 5 |
*/ |
| 6 |
|
| 7 |
#include <linux/blkdev.h> |
| 8 |
#include <linux/fs.h> |
| 9 |
#include <linux/mm.h> |
| 10 |
#include <linux/module.h> |
| 11 |
#include <linux/bio.h> |
| 12 |
#include <linux/slab.h> |
| 13 |
#include <linux/init.h> |
| 14 |
#include <linux/buffer_head.h> |
| 15 |
#include <linux/mempool.h> |
| 16 |
#include <linux/seq_file.h> |
| 17 |
#include <linux/writeback.h> |
| 18 |
#include <linux/migrate.h> |
| 19 |
#include "jfs_incore.h" |
| 20 |
#include "jfs_superblock.h" |
| 21 |
#include "jfs_filsys.h" |
| 22 |
#include "jfs_metapage.h" |
| 23 |
#include "jfs_txnmgr.h" |
| 24 |
#include "jfs_debug.h" |
| 25 |
|
| 26 |
#ifdef CONFIG_JFS_STATISTICS |
| 27 |
static struct { |
| 28 |
uint pagealloc; /* # of page allocations */ |
| 29 |
uint pagefree; /* # of page frees */ |
| 30 |
uint lockwait; /* # of sleeping lock_metapage() calls */ |
| 31 |
} mpStat; |
| 32 |
#endif |
| 33 |
|
| 34 |
#define metapage_locked(mp) test_bit(META_locked, &(mp)->flag) |
| 35 |
#define trylock_metapage(mp) test_and_set_bit_lock(META_locked, &(mp)->flag) |
| 36 |
|
| 37 |
static inline void unlock_metapage(struct metapage *mp) |
| 38 |
{ |
| 39 |
clear_bit_unlock(META_locked, &mp->flag); |
| 40 |
wake_up(&mp->wait); |
| 41 |
} |
| 42 |
|
| 43 |
static inline void __lock_metapage(struct metapage *mp) |
| 44 |
{ |
| 45 |
DECLARE_WAITQUEUE(wait, current); |
| 46 |
INCREMENT(mpStat.lockwait); |
| 47 |
add_wait_queue_exclusive(&mp->wait, &wait); |
| 48 |
do { |
| 49 |
set_current_state(TASK_UNINTERRUPTIBLE); |
| 50 |
if (metapage_locked(mp)) { |
| 51 |
folio_unlock(mp->folio); |
| 52 |
io_schedule(); |
| 53 |
folio_lock(mp->folio); |
| 54 |
} |
| 55 |
} while (trylock_metapage(mp)); |
| 56 |
__set_current_state(TASK_RUNNING); |
| 57 |
remove_wait_queue(&mp->wait, &wait); |
| 58 |
} |
| 59 |
|
| 60 |
/* |
| 61 |
* Must have mp->folio locked |
| 62 |
*/ |
| 63 |
static inline void lock_metapage(struct metapage *mp) |
| 64 |
{ |
| 65 |
if (trylock_metapage(mp)) |
| 66 |
__lock_metapage(mp); |
| 67 |
} |
| 68 |
|
| 69 |
#define METAPOOL_MIN_PAGES 32 |
| 70 |
static struct kmem_cache *metapage_cache; |
| 71 |
static mempool_t *metapage_mempool; |
| 72 |
|
| 73 |
#define MPS_PER_PAGE (PAGE_SIZE >> L2PSIZE) |
| 74 |
|
| 75 |
#if MPS_PER_PAGE > 1 |
| 76 |
|
| 77 |
struct meta_anchor { |
| 78 |
int mp_count; |
| 79 |
atomic_t io_count; |
| 80 |
blk_status_t status; |
| 81 |
struct metapage *mp[MPS_PER_PAGE]; |
| 82 |
}; |
| 83 |
|
| 84 |
static inline struct metapage *folio_to_mp(struct folio *folio, int offset) |
| 85 |
{ |
| 86 |
struct meta_anchor *anchor = folio->private; |
| 87 |
|
| 88 |
if (!anchor) |
| 89 |
return NULL; |
| 90 |
return anchor->mp[offset >> L2PSIZE]; |
| 91 |
} |
| 92 |
|
| 93 |
static inline int insert_metapage(struct folio *folio, struct metapage *mp) |
| 94 |
{ |
| 95 |
struct meta_anchor *a; |
| 96 |
int index; |
| 97 |
int l2mp_blocks; /* log2 blocks per metapage */ |
| 98 |
|
| 99 |
a = folio->private; |
| 100 |
if (!a) { |
| 101 |
a = kzalloc_obj(struct meta_anchor, GFP_NOFS); |
| 102 |
if (!a) |
| 103 |
return -ENOMEM; |
| 104 |
folio_attach_private(folio, a); |
| 105 |
kmap(&folio->page); |
| 106 |
} |
| 107 |
|
| 108 |
if (mp) { |
| 109 |
l2mp_blocks = L2PSIZE - folio->mapping->host->i_blkbits; |
| 110 |
index = (mp->index >> l2mp_blocks) & (MPS_PER_PAGE - 1); |
| 111 |
a->mp_count++; |
| 112 |
a->mp[index] = mp; |
| 113 |
} |
| 114 |
|
| 115 |
return 0; |
| 116 |
} |
| 117 |
|
| 118 |
static inline void remove_metapage(struct folio *folio, struct metapage *mp) |
| 119 |
{ |
| 120 |
struct meta_anchor *a = folio->private; |
| 121 |
int l2mp_blocks = L2PSIZE - folio->mapping->host->i_blkbits; |
| 122 |
int index; |
| 123 |
|
| 124 |
index = (mp->index >> l2mp_blocks) & (MPS_PER_PAGE - 1); |
| 125 |
|
| 126 |
BUG_ON(a->mp[index] != mp); |
| 127 |
|
| 128 |
a->mp[index] = NULL; |
| 129 |
if (--a->mp_count == 0) { |
| 130 |
kfree(a); |
| 131 |
folio_detach_private(folio); |
| 132 |
kunmap(&folio->page); |
| 133 |
} |
| 134 |
} |
| 135 |
|
| 136 |
static inline void inc_io(struct folio *folio) |
| 137 |
{ |
| 138 |
struct meta_anchor *anchor = folio->private; |
| 139 |
|
| 140 |
atomic_inc(&anchor->io_count); |
| 141 |
} |
| 142 |
|
| 143 |
static inline void dec_io(struct folio *folio, blk_status_t status, |
| 144 |
void (*handler)(struct folio *, blk_status_t)) |
| 145 |
{ |
| 146 |
struct meta_anchor *anchor = folio->private; |
| 147 |
|
| 148 |
if (anchor->status == BLK_STS_OK) |
| 149 |
anchor->status = status; |
| 150 |
|
| 151 |
if (atomic_dec_and_test(&anchor->io_count)) |
| 152 |
handler(folio, anchor->status); |
| 153 |
} |
| 154 |
|
| 155 |
#ifdef CONFIG_MIGRATION |
| 156 |
static int __metapage_migrate_folio(struct address_space *mapping, |
| 157 |
struct folio *dst, struct folio *src, |
| 158 |
enum migrate_mode mode) |
| 159 |
{ |
| 160 |
struct meta_anchor *src_anchor = src->private; |
| 161 |
struct metapage *mps[MPS_PER_PAGE] = {0}; |
| 162 |
struct metapage *mp; |
| 163 |
int i, rc; |
| 164 |
|
| 165 |
for (i = 0; i < MPS_PER_PAGE; i++) { |
| 166 |
mp = src_anchor->mp[i]; |
| 167 |
if (mp && metapage_locked(mp)) |
| 168 |
return -EAGAIN; |
| 169 |
} |
| 170 |
|
| 171 |
rc = filemap_migrate_folio(mapping, dst, src, mode); |
| 172 |
if (rc) |
| 173 |
return rc; |
| 174 |
|
| 175 |
for (i = 0; i < MPS_PER_PAGE; i++) { |
| 176 |
mp = src_anchor->mp[i]; |
| 177 |
if (!mp) |
| 178 |
continue; |
| 179 |
if (unlikely(insert_metapage(dst, mp))) { |
| 180 |
/* If error, roll-back previosly inserted pages */ |
| 181 |
for (int j = 0 ; j < i; j++) { |
| 182 |
if (mps[j]) |
| 183 |
remove_metapage(dst, mps[j]); |
| 184 |
} |
| 185 |
return -EAGAIN; |
| 186 |
} |
| 187 |
mps[i] = mp; |
| 188 |
} |
| 189 |
|
| 190 |
/* Update the metapage and remove it from src */ |
| 191 |
for (i = 0; i < MPS_PER_PAGE; i++) { |
| 192 |
mp = mps[i]; |
| 193 |
if (mp) { |
| 194 |
int page_offset = mp->data - folio_address(src); |
| 195 |
|
| 196 |
mp->data = folio_address(dst) + page_offset; |
| 197 |
mp->folio = dst; |
| 198 |
remove_metapage(src, mp); |
| 199 |
} |
| 200 |
} |
| 201 |
|
| 202 |
return 0; |
| 203 |
} |
| 204 |
#endif /* CONFIG_MIGRATION */ |
| 205 |
|
| 206 |
#else |
| 207 |
|
| 208 |
static inline struct metapage *folio_to_mp(struct folio *folio, int offset) |
| 209 |
{ |
| 210 |
return folio->private; |
| 211 |
} |
| 212 |
|
| 213 |
static inline int insert_metapage(struct folio *folio, struct metapage *mp) |
| 214 |
{ |
| 215 |
if (mp) { |
| 216 |
folio_attach_private(folio, mp); |
| 217 |
kmap(&folio->page); |
| 218 |
} |
| 219 |
return 0; |
| 220 |
} |
| 221 |
|
| 222 |
static inline void remove_metapage(struct folio *folio, struct metapage *mp) |
| 223 |
{ |
| 224 |
folio_detach_private(folio); |
| 225 |
kunmap(&folio->page); |
| 226 |
} |
| 227 |
|
| 228 |
#define inc_io(folio) do {} while(0) |
| 229 |
#define dec_io(folio, status, handler) handler(folio, status) |
| 230 |
|
| 231 |
#ifdef CONFIG_MIGRATION |
| 232 |
static int __metapage_migrate_folio(struct address_space *mapping, |
| 233 |
struct folio *dst, struct folio *src, |
| 234 |
enum migrate_mode mode) |
| 235 |
{ |
| 236 |
struct metapage *mp; |
| 237 |
int page_offset; |
| 238 |
int rc; |
| 239 |
|
| 240 |
mp = folio_to_mp(src, 0); |
| 241 |
if (metapage_locked(mp)) |
| 242 |
return -EAGAIN; |
| 243 |
|
| 244 |
rc = filemap_migrate_folio(mapping, dst, src, mode); |
| 245 |
if (rc) |
| 246 |
return rc; |
| 247 |
|
| 248 |
if (unlikely(insert_metapage(dst, mp))) |
| 249 |
return -EAGAIN; |
| 250 |
|
| 251 |
page_offset = mp->data - folio_address(src); |
| 252 |
mp->data = folio_address(dst) + page_offset; |
| 253 |
mp->folio = dst; |
| 254 |
remove_metapage(src, mp); |
| 255 |
|
| 256 |
return 0; |
| 257 |
} |
| 258 |
#endif /* CONFIG_MIGRATION */ |
| 259 |
|
| 260 |
#endif |
| 261 |
|
| 262 |
static inline struct metapage *alloc_metapage(gfp_t gfp_mask) |
| 263 |
{ |
| 264 |
struct metapage *mp = mempool_alloc(metapage_mempool, gfp_mask); |
| 265 |
|
| 266 |
if (mp) { |
| 267 |
mp->lid = 0; |
| 268 |
mp->lsn = 0; |
| 269 |
mp->data = NULL; |
| 270 |
mp->clsn = 0; |
| 271 |
mp->log = NULL; |
| 272 |
init_waitqueue_head(&mp->wait); |
| 273 |
} |
| 274 |
return mp; |
| 275 |
} |
| 276 |
|
| 277 |
static inline void free_metapage(struct metapage *mp) |
| 278 |
{ |
| 279 |
mempool_free(mp, metapage_mempool); |
| 280 |
} |
| 281 |
|
| 282 |
int __init metapage_init(void) |
| 283 |
{ |
| 284 |
/* |
| 285 |
* Allocate the metapage structures |
| 286 |
*/ |
| 287 |
metapage_cache = kmem_cache_create("jfs_mp", sizeof(struct metapage), |
| 288 |
0, 0, NULL); |
| 289 |
if (metapage_cache == NULL) |
| 290 |
return -ENOMEM; |
| 291 |
|
| 292 |
metapage_mempool = mempool_create_slab_pool(METAPOOL_MIN_PAGES, |
| 293 |
metapage_cache); |
| 294 |
|
| 295 |
if (metapage_mempool == NULL) { |
| 296 |
kmem_cache_destroy(metapage_cache); |
| 297 |
return -ENOMEM; |
| 298 |
} |
| 299 |
|
| 300 |
return 0; |
| 301 |
} |
| 302 |
|
| 303 |
void metapage_exit(void) |
| 304 |
{ |
| 305 |
mempool_destroy(metapage_mempool); |
| 306 |
kmem_cache_destroy(metapage_cache); |
| 307 |
} |
| 308 |
|
| 309 |
static inline void drop_metapage(struct folio *folio, struct metapage *mp) |
| 310 |
{ |
| 311 |
if (mp->count || mp->nohomeok || test_bit(META_dirty, &mp->flag) || |
| 312 |
test_bit(META_io, &mp->flag)) |
| 313 |
return; |
| 314 |
remove_metapage(folio, mp); |
| 315 |
INCREMENT(mpStat.pagefree); |
| 316 |
free_metapage(mp); |
| 317 |
} |
| 318 |
|
| 319 |
/* |
| 320 |
* Metapage address space operations |
| 321 |
*/ |
| 322 |
|
| 323 |
static sector_t metapage_get_blocks(struct inode *inode, sector_t lblock, |
| 324 |
int *len) |
| 325 |
{ |
| 326 |
int rc = 0; |
| 327 |
int xflag; |
| 328 |
s64 xaddr; |
| 329 |
sector_t file_blocks = (inode->i_size + inode->i_sb->s_blocksize - 1) >> |
| 330 |
inode->i_blkbits; |
| 331 |
|
| 332 |
if (lblock >= file_blocks) |
| 333 |
return 0; |
| 334 |
if (lblock + *len > file_blocks) |
| 335 |
*len = file_blocks - lblock; |
| 336 |
|
| 337 |
if (inode->i_ino) { |
| 338 |
rc = xtLookup(inode, (s64)lblock, *len, &xflag, &xaddr, len, 0); |
| 339 |
if ((rc == 0) && *len) |
| 340 |
lblock = (sector_t)xaddr; |
| 341 |
else |
| 342 |
lblock = 0; |
| 343 |
} /* else no mapping */ |
| 344 |
|
| 345 |
return lblock; |
| 346 |
} |
| 347 |
|
| 348 |
static void last_read_complete(struct folio *folio, blk_status_t status) |
| 349 |
{ |
| 350 |
if (status) |
| 351 |
printk(KERN_ERR "Read error %d at %#llx\n", status, |
| 352 |
folio_pos(folio)); |
| 353 |
|
| 354 |
folio_end_read(folio, status == 0); |
| 355 |
} |
| 356 |
|
| 357 |
static void metapage_read_end_io(struct bio *bio) |
| 358 |
{ |
| 359 |
struct folio *folio = bio->bi_private; |
| 360 |
|
| 361 |
dec_io(folio, bio->bi_status, last_read_complete); |
| 362 |
bio_put(bio); |
| 363 |
} |
| 364 |
|
| 365 |
static void remove_from_logsync(struct metapage *mp) |
| 366 |
{ |
| 367 |
struct jfs_log *log = mp->log; |
| 368 |
unsigned long flags; |
| 369 |
/* |
| 370 |
* This can race. Recheck that log hasn't been set to null, and after |
| 371 |
* acquiring logsync lock, recheck lsn |
| 372 |
*/ |
| 373 |
if (!log) |
| 374 |
return; |
| 375 |
|
| 376 |
LOGSYNC_LOCK(log, flags); |
| 377 |
if (mp->lsn) { |
| 378 |
mp->log = NULL; |
| 379 |
mp->lsn = 0; |
| 380 |
mp->clsn = 0; |
| 381 |
log->count--; |
| 382 |
list_del(&mp->synclist); |
| 383 |
} |
| 384 |
LOGSYNC_UNLOCK(log, flags); |
| 385 |
} |
| 386 |
|
| 387 |
static void last_write_complete(struct folio *folio, blk_status_t status) |
| 388 |
{ |
| 389 |
struct metapage *mp; |
| 390 |
unsigned int offset; |
| 391 |
|
| 392 |
if (status) { |
| 393 |
int err = blk_status_to_errno(status); |
| 394 |
printk(KERN_ERR "metapage_write_end_io: I/O error\n"); |
| 395 |
mapping_set_error(folio->mapping, err); |
| 396 |
} |
| 397 |
|
| 398 |
for (offset = 0; offset < PAGE_SIZE; offset += PSIZE) { |
| 399 |
mp = folio_to_mp(folio, offset); |
| 400 |
if (mp && test_bit(META_io, &mp->flag)) { |
| 401 |
if (mp->lsn) |
| 402 |
remove_from_logsync(mp); |
| 403 |
clear_bit(META_io, &mp->flag); |
| 404 |
} |
| 405 |
/* |
| 406 |
* I'd like to call drop_metapage here, but I don't think it's |
| 407 |
* safe unless I have the page locked |
| 408 |
*/ |
| 409 |
} |
| 410 |
folio_end_writeback(folio); |
| 411 |
} |
| 412 |
|
| 413 |
static void metapage_write_end_io(struct bio *bio) |
| 414 |
{ |
| 415 |
struct folio *folio = bio->bi_private; |
| 416 |
|
| 417 |
BUG_ON(!folio->private); |
| 418 |
|
| 419 |
dec_io(folio, bio->bi_status, last_write_complete); |
| 420 |
bio_put(bio); |
| 421 |
} |
| 422 |
|
| 423 |
static int metapage_write_folio(struct folio *folio, |
| 424 |
struct writeback_control *wbc) |
| 425 |
{ |
| 426 |
struct bio *bio = NULL; |
| 427 |
int block_offset; /* block offset of mp within page */ |
| 428 |
struct inode *inode = folio->mapping->host; |
| 429 |
int blocks_per_mp = JFS_SBI(inode->i_sb)->nbperpage; |
| 430 |
int len; |
| 431 |
int xlen; |
| 432 |
struct metapage *mp; |
| 433 |
int redirty = 0; |
| 434 |
sector_t lblock; |
| 435 |
int nr_underway = 0; |
| 436 |
sector_t pblock; |
| 437 |
sector_t next_block = 0; |
| 438 |
sector_t page_start; |
| 439 |
unsigned long bio_bytes = 0; |
| 440 |
unsigned long bio_offset = 0; |
| 441 |
int offset; |
| 442 |
int bad_blocks = 0; |
| 443 |
|
| 444 |
page_start = folio_pos(folio) >> inode->i_blkbits; |
| 445 |
BUG_ON(!folio_test_locked(folio)); |
| 446 |
BUG_ON(folio_test_writeback(folio)); |
| 447 |
folio_start_writeback(folio); |
| 448 |
|
| 449 |
for (offset = 0; offset < PAGE_SIZE; offset += PSIZE) { |
| 450 |
mp = folio_to_mp(folio, offset); |
| 451 |
|
| 452 |
if (!mp || !test_bit(META_dirty, &mp->flag)) |
| 453 |
continue; |
| 454 |
|
| 455 |
if (mp->nohomeok && !test_bit(META_forcewrite, &mp->flag)) { |
| 456 |
redirty = 1; |
| 457 |
/* |
| 458 |
* Make sure this page isn't blocked indefinitely. |
| 459 |
* If the journal isn't undergoing I/O, push it |
| 460 |
*/ |
| 461 |
if (mp->log && !(mp->log->cflag & logGC_PAGEOUT)) |
| 462 |
jfs_flush_journal(mp->log, 0); |
| 463 |
continue; |
| 464 |
} |
| 465 |
|
| 466 |
clear_bit(META_dirty, &mp->flag); |
| 467 |
set_bit(META_io, &mp->flag); |
| 468 |
block_offset = offset >> inode->i_blkbits; |
| 469 |
lblock = page_start + block_offset; |
| 470 |
if (bio) { |
| 471 |
if (xlen && lblock == next_block) { |
| 472 |
/* Contiguous, in memory & on disk */ |
| 473 |
len = min(xlen, blocks_per_mp); |
| 474 |
xlen -= len; |
| 475 |
bio_bytes += len << inode->i_blkbits; |
| 476 |
continue; |
| 477 |
} |
| 478 |
/* Not contiguous */ |
| 479 |
bio_add_folio_nofail(bio, folio, bio_bytes, bio_offset); |
| 480 |
/* |
| 481 |
* Increment counter before submitting i/o to keep |
| 482 |
* count from hitting zero before we're through |
| 483 |
*/ |
| 484 |
inc_io(folio); |
| 485 |
if (!bio->bi_iter.bi_size) |
| 486 |
goto dump_bio; |
| 487 |
submit_bio(bio); |
| 488 |
nr_underway++; |
| 489 |
bio = NULL; |
| 490 |
} else |
| 491 |
inc_io(folio); |
| 492 |
xlen = (folio_size(folio) - offset) >> inode->i_blkbits; |
| 493 |
pblock = metapage_get_blocks(inode, lblock, &xlen); |
| 494 |
if (!pblock) { |
| 495 |
printk(KERN_ERR "JFS: metapage_get_blocks failed\n"); |
| 496 |
/* |
| 497 |
* We already called inc_io(), but can't cancel it |
| 498 |
* with dec_io() until we're done with the page |
| 499 |
*/ |
| 500 |
bad_blocks++; |
| 501 |
continue; |
| 502 |
} |
| 503 |
len = min(xlen, (int)JFS_SBI(inode->i_sb)->nbperpage); |
| 504 |
|
| 505 |
bio = bio_alloc(inode->i_sb->s_bdev, 1, REQ_OP_WRITE, GFP_NOFS); |
| 506 |
bio->bi_iter.bi_sector = pblock << (inode->i_blkbits - 9); |
| 507 |
bio->bi_end_io = metapage_write_end_io; |
| 508 |
bio->bi_private = folio; |
| 509 |
|
| 510 |
/* Don't call bio_add_page yet, we may add to this vec */ |
| 511 |
bio_offset = offset; |
| 512 |
bio_bytes = len << inode->i_blkbits; |
| 513 |
|
| 514 |
xlen -= len; |
| 515 |
next_block = lblock + len; |
| 516 |
} |
| 517 |
if (bio) { |
| 518 |
bio_add_folio_nofail(bio, folio, bio_bytes, bio_offset); |
| 519 |
if (!bio->bi_iter.bi_size) |
| 520 |
goto dump_bio; |
| 521 |
|
| 522 |
submit_bio(bio); |
| 523 |
nr_underway++; |
| 524 |
} |
| 525 |
if (redirty) |
| 526 |
folio_redirty_for_writepage(wbc, folio); |
| 527 |
|
| 528 |
folio_unlock(folio); |
| 529 |
|
| 530 |
if (bad_blocks) |
| 531 |
goto err_out; |
| 532 |
|
| 533 |
if (nr_underway == 0) |
| 534 |
folio_end_writeback(folio); |
| 535 |
|
| 536 |
return 0; |
| 537 |
dump_bio: |
| 538 |
print_hex_dump(KERN_ERR, "JFS: dump of bio: ", DUMP_PREFIX_ADDRESS, 16, |
| 539 |
4, bio, sizeof(*bio), 0); |
| 540 |
bio_put(bio); |
| 541 |
folio_unlock(folio); |
| 542 |
dec_io(folio, BLK_STS_OK, last_write_complete); |
| 543 |
err_out: |
| 544 |
while (bad_blocks--) |
| 545 |
dec_io(folio, BLK_STS_OK, last_write_complete); |
| 546 |
return -EIO; |
| 547 |
} |
| 548 |
|
| 549 |
static int metapage_writepages(struct address_space *mapping, |
| 550 |
struct writeback_control *wbc) |
| 551 |
{ |
| 552 |
struct blk_plug plug; |
| 553 |
struct folio *folio = NULL; |
| 554 |
int err; |
| 555 |
|
| 556 |
blk_start_plug(&plug); |
| 557 |
while ((folio = writeback_iter(mapping, wbc, folio, &err))) |
| 558 |
err = metapage_write_folio(folio, wbc); |
| 559 |
blk_finish_plug(&plug); |
| 560 |
|
| 561 |
return err; |
| 562 |
} |
| 563 |
|
| 564 |
static int metapage_read_folio(struct file *fp, struct folio *folio) |
| 565 |
{ |
| 566 |
struct inode *inode = folio->mapping->host; |
| 567 |
struct bio *bio = NULL; |
| 568 |
int block_offset; |
| 569 |
int blocks_per_page = i_blocks_per_folio(inode, folio); |
| 570 |
sector_t page_start; /* address of page in fs blocks */ |
| 571 |
sector_t pblock; |
| 572 |
int xlen; |
| 573 |
unsigned int len; |
| 574 |
int offset; |
| 575 |
|
| 576 |
BUG_ON(!folio_test_locked(folio)); |
| 577 |
page_start = folio_pos(folio) >> inode->i_blkbits; |
| 578 |
|
| 579 |
block_offset = 0; |
| 580 |
while (block_offset < blocks_per_page) { |
| 581 |
xlen = blocks_per_page - block_offset; |
| 582 |
pblock = metapage_get_blocks(inode, page_start + block_offset, |
| 583 |
&xlen); |
| 584 |
if (pblock) { |
| 585 |
if (!folio->private) |
| 586 |
insert_metapage(folio, NULL); |
| 587 |
inc_io(folio); |
| 588 |
if (bio) |
| 589 |
submit_bio(bio); |
| 590 |
|
| 591 |
bio = bio_alloc(inode->i_sb->s_bdev, 1, REQ_OP_READ, |
| 592 |
GFP_NOFS); |
| 593 |
bio->bi_iter.bi_sector = |
| 594 |
pblock << (inode->i_blkbits - 9); |
| 595 |
bio->bi_end_io = metapage_read_end_io; |
| 596 |
bio->bi_private = folio; |
| 597 |
len = xlen << inode->i_blkbits; |
| 598 |
offset = block_offset << inode->i_blkbits; |
| 599 |
bio_add_folio_nofail(bio, folio, len, offset); |
| 600 |
block_offset += xlen; |
| 601 |
} else |
| 602 |
block_offset++; |
| 603 |
} |
| 604 |
if (bio) |
| 605 |
submit_bio(bio); |
| 606 |
else |
| 607 |
folio_unlock(folio); |
| 608 |
|
| 609 |
return 0; |
| 610 |
} |
| 611 |
|
| 612 |
static bool metapage_release_folio(struct folio *folio, gfp_t gfp_mask) |
| 613 |
{ |
| 614 |
struct metapage *mp; |
| 615 |
bool ret = true; |
| 616 |
int offset; |
| 617 |
|
| 618 |
for (offset = 0; offset < PAGE_SIZE; offset += PSIZE) { |
| 619 |
mp = folio_to_mp(folio, offset); |
| 620 |
|
| 621 |
if (!mp) |
| 622 |
continue; |
| 623 |
|
| 624 |
jfs_info("metapage_release_folio: mp = 0x%p", mp); |
| 625 |
if (mp->count || mp->nohomeok || |
| 626 |
test_bit(META_dirty, &mp->flag)) { |
| 627 |
jfs_info("count = %ld, nohomeok = %d", mp->count, |
| 628 |
mp->nohomeok); |
| 629 |
ret = false; |
| 630 |
continue; |
| 631 |
} |
| 632 |
if (mp->lsn) |
| 633 |
remove_from_logsync(mp); |
| 634 |
remove_metapage(folio, mp); |
| 635 |
INCREMENT(mpStat.pagefree); |
| 636 |
free_metapage(mp); |
| 637 |
} |
| 638 |
return ret; |
| 639 |
} |
| 640 |
|
| 641 |
#ifdef CONFIG_MIGRATION |
| 642 |
/* |
| 643 |
* metapage_migrate_folio - Migration function for JFS metapages |
| 644 |
*/ |
| 645 |
static int metapage_migrate_folio(struct address_space *mapping, |
| 646 |
struct folio *dst, struct folio *src, |
| 647 |
enum migrate_mode mode) |
| 648 |
{ |
| 649 |
int expected_count; |
| 650 |
|
| 651 |
if (!src->private) |
| 652 |
return filemap_migrate_folio(mapping, dst, src, mode); |
| 653 |
|
| 654 |
/* Check whether page does not have extra refs before we do more work */ |
| 655 |
expected_count = folio_expected_ref_count(src) + 1; |
| 656 |
if (folio_ref_count(src) != expected_count) |
| 657 |
return -EAGAIN; |
| 658 |
return __metapage_migrate_folio(mapping, dst, src, mode); |
| 659 |
} |
| 660 |
#else |
| 661 |
#define metapage_migrate_folio NULL |
| 662 |
#endif /* CONFIG_MIGRATION */ |
| 663 |
|
| 664 |
static void metapage_invalidate_folio(struct folio *folio, size_t offset, |
| 665 |
size_t length) |
| 666 |
{ |
| 667 |
BUG_ON(offset || length < folio_size(folio)); |
| 668 |
|
| 669 |
BUG_ON(folio_test_writeback(folio)); |
| 670 |
|
| 671 |
metapage_release_folio(folio, 0); |
| 672 |
} |
| 673 |
|
| 674 |
const struct address_space_operations jfs_metapage_aops = { |
| 675 |
.read_folio = metapage_read_folio, |
| 676 |
.writepages = metapage_writepages, |
| 677 |
.release_folio = metapage_release_folio, |
| 678 |
.invalidate_folio = metapage_invalidate_folio, |
| 679 |
.dirty_folio = filemap_dirty_folio, |
| 680 |
.migrate_folio = metapage_migrate_folio, |
| 681 |
}; |
| 682 |
|
| 683 |
struct metapage *__get_metapage(struct inode *inode, unsigned long lblock, |
| 684 |
unsigned int size, int absolute, |
| 685 |
unsigned long new) |
| 686 |
{ |
| 687 |
int l2BlocksPerPage; |
| 688 |
int l2bsize; |
| 689 |
struct address_space *mapping; |
| 690 |
struct metapage *mp = NULL; |
| 691 |
struct folio *folio; |
| 692 |
unsigned long page_index; |
| 693 |
unsigned long page_offset; |
| 694 |
|
| 695 |
jfs_info("__get_metapage: ino = %ld, lblock = 0x%lx, abs=%d", |
| 696 |
inode->i_ino, lblock, absolute); |
| 697 |
|
| 698 |
l2bsize = inode->i_blkbits; |
| 699 |
l2BlocksPerPage = PAGE_SHIFT - l2bsize; |
| 700 |
page_index = lblock >> l2BlocksPerPage; |
| 701 |
page_offset = (lblock - (page_index << l2BlocksPerPage)) << l2bsize; |
| 702 |
if ((page_offset + size) > PAGE_SIZE) { |
| 703 |
jfs_err("MetaData crosses page boundary!!"); |
| 704 |
jfs_err("lblock = %lx, size = %d", lblock, size); |
| 705 |
dump_stack(); |
| 706 |
return NULL; |
| 707 |
} |
| 708 |
if (absolute) |
| 709 |
mapping = JFS_SBI(inode->i_sb)->direct_inode->i_mapping; |
| 710 |
else { |
| 711 |
/* |
| 712 |
* If an nfs client tries to read an inode that is larger |
| 713 |
* than any existing inodes, we may try to read past the |
| 714 |
* end of the inode map |
| 715 |
*/ |
| 716 |
if ((lblock << inode->i_blkbits) >= inode->i_size) |
| 717 |
return NULL; |
| 718 |
mapping = inode->i_mapping; |
| 719 |
} |
| 720 |
|
| 721 |
if (new && (PSIZE == PAGE_SIZE)) { |
| 722 |
folio = filemap_grab_folio(mapping, page_index); |
| 723 |
if (IS_ERR(folio)) { |
| 724 |
jfs_err("filemap_grab_folio failed!"); |
| 725 |
return NULL; |
| 726 |
} |
| 727 |
folio_mark_uptodate(folio); |
| 728 |
} else { |
| 729 |
folio = read_mapping_folio(mapping, page_index, NULL); |
| 730 |
if (IS_ERR(folio)) { |
| 731 |
jfs_err("read_mapping_page failed!"); |
| 732 |
return NULL; |
| 733 |
} |
| 734 |
folio_lock(folio); |
| 735 |
} |
| 736 |
|
| 737 |
mp = folio_to_mp(folio, page_offset); |
| 738 |
if (mp) { |
| 739 |
if (mp->logical_size != size) { |
| 740 |
jfs_error(inode->i_sb, |
| 741 |
"get_mp->logical_size != size\n"); |
| 742 |
jfs_err("logical_size = %d, size = %d", |
| 743 |
mp->logical_size, size); |
| 744 |
dump_stack(); |
| 745 |
goto unlock; |
| 746 |
} |
| 747 |
mp->count++; |
| 748 |
lock_metapage(mp); |
| 749 |
if (test_bit(META_discard, &mp->flag)) { |
| 750 |
if (!new) { |
| 751 |
jfs_error(inode->i_sb, |
| 752 |
"using a discarded metapage\n"); |
| 753 |
discard_metapage(mp); |
| 754 |
goto unlock; |
| 755 |
} |
| 756 |
clear_bit(META_discard, &mp->flag); |
| 757 |
} |
| 758 |
} else { |
| 759 |
INCREMENT(mpStat.pagealloc); |
| 760 |
mp = alloc_metapage(GFP_NOFS); |
| 761 |
if (!mp) |
| 762 |
goto unlock; |
| 763 |
mp->folio = folio; |
| 764 |
mp->sb = inode->i_sb; |
| 765 |
mp->flag = 0; |
| 766 |
mp->xflag = COMMIT_PAGE; |
| 767 |
mp->count = 1; |
| 768 |
mp->nohomeok = 0; |
| 769 |
mp->logical_size = size; |
| 770 |
mp->data = folio_address(folio) + page_offset; |
| 771 |
mp->index = lblock; |
| 772 |
if (unlikely(insert_metapage(folio, mp))) { |
| 773 |
free_metapage(mp); |
| 774 |
goto unlock; |
| 775 |
} |
| 776 |
lock_metapage(mp); |
| 777 |
} |
| 778 |
|
| 779 |
if (new) { |
| 780 |
jfs_info("zeroing mp = 0x%p", mp); |
| 781 |
memset(mp->data, 0, PSIZE); |
| 782 |
} |
| 783 |
|
| 784 |
folio_unlock(folio); |
| 785 |
jfs_info("__get_metapage: returning = 0x%p data = 0x%p", mp, mp->data); |
| 786 |
return mp; |
| 787 |
|
| 788 |
unlock: |
| 789 |
folio_unlock(folio); |
| 790 |
return NULL; |
| 791 |
} |
| 792 |
|
| 793 |
void grab_metapage(struct metapage * mp) |
| 794 |
{ |
| 795 |
jfs_info("grab_metapage: mp = 0x%p", mp); |
| 796 |
folio_get(mp->folio); |
| 797 |
folio_lock(mp->folio); |
| 798 |
mp->count++; |
| 799 |
lock_metapage(mp); |
| 800 |
folio_unlock(mp->folio); |
| 801 |
} |
| 802 |
|
| 803 |
static int metapage_write_one(struct folio *folio) |
| 804 |
{ |
| 805 |
struct address_space *mapping = folio->mapping; |
| 806 |
struct writeback_control wbc = { |
| 807 |
.sync_mode = WB_SYNC_ALL, |
| 808 |
.nr_to_write = folio_nr_pages(folio), |
| 809 |
}; |
| 810 |
int ret = 0; |
| 811 |
|
| 812 |
BUG_ON(!folio_test_locked(folio)); |
| 813 |
|
| 814 |
folio_wait_writeback(folio); |
| 815 |
|
| 816 |
if (folio_clear_dirty_for_io(folio)) { |
| 817 |
folio_get(folio); |
| 818 |
ret = metapage_write_folio(folio, &wbc); |
| 819 |
if (ret == 0) |
| 820 |
folio_wait_writeback(folio); |
| 821 |
folio_put(folio); |
| 822 |
} else { |
| 823 |
folio_unlock(folio); |
| 824 |
} |
| 825 |
|
| 826 |
if (!ret) |
| 827 |
ret = filemap_check_errors(mapping); |
| 828 |
return ret; |
| 829 |
} |
| 830 |
|
| 831 |
void force_metapage(struct metapage *mp) |
| 832 |
{ |
| 833 |
struct folio *folio = mp->folio; |
| 834 |
jfs_info("force_metapage: mp = 0x%p", mp); |
| 835 |
set_bit(META_forcewrite, &mp->flag); |
| 836 |
clear_bit(META_sync, &mp->flag); |
| 837 |
folio_get(folio); |
| 838 |
folio_lock(folio); |
| 839 |
folio_mark_dirty(folio); |
| 840 |
if (metapage_write_one(folio)) |
| 841 |
jfs_error(mp->sb, "metapage_write_one() failed\n"); |
| 842 |
clear_bit(META_forcewrite, &mp->flag); |
| 843 |
folio_put(folio); |
| 844 |
} |
| 845 |
|
| 846 |
void hold_metapage(struct metapage *mp) |
| 847 |
{ |
| 848 |
folio_lock(mp->folio); |
| 849 |
} |
| 850 |
|
| 851 |
void put_metapage(struct metapage *mp) |
| 852 |
{ |
| 853 |
if (mp->count || mp->nohomeok) { |
| 854 |
/* Someone else will release this */ |
| 855 |
folio_unlock(mp->folio); |
| 856 |
return; |
| 857 |
} |
| 858 |
folio_get(mp->folio); |
| 859 |
mp->count++; |
| 860 |
lock_metapage(mp); |
| 861 |
folio_unlock(mp->folio); |
| 862 |
release_metapage(mp); |
| 863 |
} |
| 864 |
|
| 865 |
void release_metapage(struct metapage * mp) |
| 866 |
{ |
| 867 |
struct folio *folio = mp->folio; |
| 868 |
jfs_info("release_metapage: mp = 0x%p, flag = 0x%lx", mp, mp->flag); |
| 869 |
|
| 870 |
folio_lock(folio); |
| 871 |
unlock_metapage(mp); |
| 872 |
|
| 873 |
assert(mp->count); |
| 874 |
if (--mp->count || mp->nohomeok) { |
| 875 |
folio_unlock(folio); |
| 876 |
folio_put(folio); |
| 877 |
return; |
| 878 |
} |
| 879 |
|
| 880 |
if (test_bit(META_dirty, &mp->flag)) { |
| 881 |
folio_mark_dirty(folio); |
| 882 |
if (test_bit(META_sync, &mp->flag)) { |
| 883 |
clear_bit(META_sync, &mp->flag); |
| 884 |
if (metapage_write_one(folio)) |
| 885 |
jfs_error(mp->sb, "metapage_write_one() failed\n"); |
| 886 |
folio_lock(folio); |
| 887 |
} |
| 888 |
} else if (mp->lsn) /* discard_metapage doesn't remove it */ |
| 889 |
remove_from_logsync(mp); |
| 890 |
|
| 891 |
/* Try to keep metapages from using up too much memory */ |
| 892 |
drop_metapage(folio, mp); |
| 893 |
|
| 894 |
folio_unlock(folio); |
| 895 |
folio_put(folio); |
| 896 |
} |
| 897 |
|
| 898 |
void __invalidate_metapages(struct inode *ip, s64 addr, int len) |
| 899 |
{ |
| 900 |
sector_t lblock; |
| 901 |
int l2BlocksPerPage = PAGE_SHIFT - ip->i_blkbits; |
| 902 |
int BlocksPerPage = 1 << l2BlocksPerPage; |
| 903 |
/* All callers are interested in block device's mapping */ |
| 904 |
struct address_space *mapping = |
| 905 |
JFS_SBI(ip->i_sb)->direct_inode->i_mapping; |
| 906 |
struct metapage *mp; |
| 907 |
unsigned int offset; |
| 908 |
|
| 909 |
/* |
| 910 |
* Mark metapages to discard. They will eventually be |
| 911 |
* released, but should not be written. |
| 912 |
*/ |
| 913 |
for (lblock = addr & ~(BlocksPerPage - 1); lblock < addr + len; |
| 914 |
lblock += BlocksPerPage) { |
| 915 |
struct folio *folio = filemap_lock_folio(mapping, |
| 916 |
lblock >> l2BlocksPerPage); |
| 917 |
if (IS_ERR(folio)) |
| 918 |
continue; |
| 919 |
for (offset = 0; offset < PAGE_SIZE; offset += PSIZE) { |
| 920 |
mp = folio_to_mp(folio, offset); |
| 921 |
if (!mp) |
| 922 |
continue; |
| 923 |
if (mp->index < addr) |
| 924 |
continue; |
| 925 |
if (mp->index >= addr + len) |
| 926 |
break; |
| 927 |
|
| 928 |
clear_bit(META_dirty, &mp->flag); |
| 929 |
set_bit(META_discard, &mp->flag); |
| 930 |
if (mp->lsn) |
| 931 |
remove_from_logsync(mp); |
| 932 |
} |
| 933 |
folio_unlock(folio); |
| 934 |
folio_put(folio); |
| 935 |
} |
| 936 |
} |
| 937 |
|
| 938 |
#ifdef CONFIG_JFS_STATISTICS |
| 939 |
int jfs_mpstat_proc_show(struct seq_file *m, void *v) |
| 940 |
{ |
| 941 |
seq_printf(m, |
| 942 |
"JFS Metapage statistics\n" |
| 943 |
"=======================\n" |
| 944 |
"page allocations = %d\n" |
| 945 |
"page frees = %d\n" |
| 946 |
"lock waits = %d\n", |
| 947 |
mpStat.pagealloc, |
| 948 |
mpStat.pagefree, |
| 949 |
mpStat.lockwait); |
| 950 |
return 0; |
| 951 |
} |
| 952 |
#endif |