1 /* 2 * Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved. 3 * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved. 4 * 5 * This copyrighted material is made available to anyone wishing to use, 6 * modify, copy, or redistribute it subject to the terms and conditions 7 * of the GNU General Public License version 2. 8 */ 9 10 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt 11 12 #include <linux/slab.h> 13 #include <linux/spinlock.h> 14 #include <linux/completion.h> 15 #include <linux/buffer_head.h> 16 #include <linux/fs.h> 17 #include <linux/gfs2_ondisk.h> 18 #include <linux/prefetch.h> 19 #include <linux/blkdev.h> 20 #include <linux/rbtree.h> 21 #include <linux/random.h> 22 23 #include "gfs2.h" 24 #include "incore.h" 25 #include "glock.h" 26 #include "glops.h" 27 #include "lops.h" 28 #include "meta_io.h" 29 #include "quota.h" 30 #include "rgrp.h" 31 #include "super.h" 32 #include "trans.h" 33 #include "util.h" 34 #include "log.h" 35 #include "inode.h" 36 #include "trace_gfs2.h" 37 #include "dir.h" 38 39 #define BFITNOENT ((u32)~0) 40 #define NO_BLOCK ((u64)~0) 41 42 #if BITS_PER_LONG == 32 43 #define LBITMASK (0x55555555UL) 44 #define LBITSKIP55 (0x55555555UL) 45 #define LBITSKIP00 (0x00000000UL) 46 #else 47 #define LBITMASK (0x5555555555555555UL) 48 #define LBITSKIP55 (0x5555555555555555UL) 49 #define LBITSKIP00 (0x0000000000000000UL) 50 #endif 51 52 /* 53 * These routines are used by the resource group routines (rgrp.c) 54 * to keep track of block allocation. Each block is represented by two 55 * bits. So, each byte represents GFS2_NBBY (i.e. 4) blocks. 56 * 57 * 0 = Free 58 * 1 = Used (not metadata) 59 * 2 = Unlinked (still in use) inode 60 * 3 = Used (metadata) 61 */ 62 63 struct gfs2_extent { 64 struct gfs2_rbm rbm; 65 u32 len; 66 }; 67 68 static const char valid_change[16] = { 69 /* current */ 70 /* n */ 0, 1, 1, 1, 71 /* e */ 1, 0, 0, 0, 72 /* w */ 0, 0, 0, 1, 73 1, 0, 0, 0 74 }; 75 76 static int gfs2_rbm_find(struct gfs2_rbm *rbm, u8 state, u32 *minext, 77 const struct gfs2_inode *ip, bool nowrap); 78 79 80 /** 81 * gfs2_setbit - Set a bit in the bitmaps 82 * @rbm: The position of the bit to set 83 * @do_clone: Also set the clone bitmap, if it exists 84 * @new_state: the new state of the block 85 * 86 */ 87 88 static inline void gfs2_setbit(const struct gfs2_rbm *rbm, bool do_clone, 89 unsigned char new_state) 90 { 91 unsigned char *byte1, *byte2, *end, cur_state; 92 struct gfs2_bitmap *bi = rbm_bi(rbm); 93 unsigned int buflen = bi->bi_bytes; 94 const unsigned int bit = (rbm->offset % GFS2_NBBY) * GFS2_BIT_SIZE; 95 96 byte1 = bi->bi_bh->b_data + bi->bi_offset + (rbm->offset / GFS2_NBBY); 97 end = bi->bi_bh->b_data + bi->bi_offset + buflen; 98 99 BUG_ON(byte1 >= end); 100 101 cur_state = (*byte1 >> bit) & GFS2_BIT_MASK; 102 103 if (unlikely(!valid_change[new_state * 4 + cur_state])) { 104 struct gfs2_sbd *sdp = rbm->rgd->rd_sbd; 105 106 fs_warn(sdp, "buf_blk = 0x%x old_state=%d, new_state=%d\n", 107 rbm->offset, cur_state, new_state); 108 fs_warn(sdp, "rgrp=0x%llx bi_start=0x%x biblk: 0x%llx\n", 109 (unsigned long long)rbm->rgd->rd_addr, bi->bi_start, 110 (unsigned long long)bi->bi_bh->b_blocknr); 111 fs_warn(sdp, "bi_offset=0x%x bi_bytes=0x%x block=0x%llx\n", 112 bi->bi_offset, bi->bi_bytes, 113 (unsigned long long)gfs2_rbm_to_block(rbm)); 114 dump_stack(); 115 gfs2_consist_rgrpd(rbm->rgd); 116 return; 117 } 118 *byte1 ^= (cur_state ^ new_state) << bit; 119 120 if (do_clone && bi->bi_clone) { 121 byte2 = bi->bi_clone + bi->bi_offset + (rbm->offset / GFS2_NBBY); 122 cur_state = (*byte2 >> bit) & GFS2_BIT_MASK; 123 *byte2 ^= (cur_state ^ new_state) << bit; 124 } 125 } 126 127 /** 128 * gfs2_testbit - test a bit in the bitmaps 129 * @rbm: The bit to test 130 * @use_clone: If true, test the clone bitmap, not the official bitmap. 131 * 132 * Some callers like gfs2_unaligned_extlen need to test the clone bitmaps, 133 * not the "real" bitmaps, to avoid allocating recently freed blocks. 134 * 135 * Returns: The two bit block state of the requested bit 136 */ 137 138 static inline u8 gfs2_testbit(const struct gfs2_rbm *rbm, bool use_clone) 139 { 140 struct gfs2_bitmap *bi = rbm_bi(rbm); 141 const u8 *buffer; 142 const u8 *byte; 143 unsigned int bit; 144 145 if (use_clone && bi->bi_clone) 146 buffer = bi->bi_clone; 147 else 148 buffer = bi->bi_bh->b_data; 149 buffer += bi->bi_offset; 150 byte = buffer + (rbm->offset / GFS2_NBBY); 151 bit = (rbm->offset % GFS2_NBBY) * GFS2_BIT_SIZE; 152 153 return (*byte >> bit) & GFS2_BIT_MASK; 154 } 155 156 /** 157 * gfs2_bit_search 158 * @ptr: Pointer to bitmap data 159 * @mask: Mask to use (normally 0x55555.... but adjusted for search start) 160 * @state: The state we are searching for 161 * 162 * We xor the bitmap data with a patter which is the bitwise opposite 163 * of what we are looking for, this gives rise to a pattern of ones 164 * wherever there is a match. Since we have two bits per entry, we 165 * take this pattern, shift it down by one place and then and it with 166 * the original. All the even bit positions (0,2,4, etc) then represent 167 * successful matches, so we mask with 0x55555..... to remove the unwanted 168 * odd bit positions. 169 * 170 * This allows searching of a whole u64 at once (32 blocks) with a 171 * single test (on 64 bit arches). 172 */ 173 174 static inline u64 gfs2_bit_search(const __le64 *ptr, u64 mask, u8 state) 175 { 176 u64 tmp; 177 static const u64 search[] = { 178 [0] = 0xffffffffffffffffULL, 179 [1] = 0xaaaaaaaaaaaaaaaaULL, 180 [2] = 0x5555555555555555ULL, 181 [3] = 0x0000000000000000ULL, 182 }; 183 tmp = le64_to_cpu(*ptr) ^ search[state]; 184 tmp &= (tmp >> 1); 185 tmp &= mask; 186 return tmp; 187 } 188 189 /** 190 * rs_cmp - multi-block reservation range compare 191 * @blk: absolute file system block number of the new reservation 192 * @len: number of blocks in the new reservation 193 * @rs: existing reservation to compare against 194 * 195 * returns: 1 if the block range is beyond the reach of the reservation 196 * -1 if the block range is before the start of the reservation 197 * 0 if the block range overlaps with the reservation 198 */ 199 static inline int rs_cmp(u64 blk, u32 len, struct gfs2_blkreserv *rs) 200 { 201 u64 startblk = gfs2_rbm_to_block(&rs->rs_rbm); 202 203 if (blk >= startblk + rs->rs_free) 204 return 1; 205 if (blk + len - 1 < startblk) 206 return -1; 207 return 0; 208 } 209 210 /** 211 * gfs2_bitfit - Search an rgrp's bitmap buffer to find a bit-pair representing 212 * a block in a given allocation state. 213 * @buf: the buffer that holds the bitmaps 214 * @len: the length (in bytes) of the buffer 215 * @goal: start search at this block's bit-pair (within @buffer) 216 * @state: GFS2_BLKST_XXX the state of the block we're looking for. 217 * 218 * Scope of @goal and returned block number is only within this bitmap buffer, 219 * not entire rgrp or filesystem. @buffer will be offset from the actual 220 * beginning of a bitmap block buffer, skipping any header structures, but 221 * headers are always a multiple of 64 bits long so that the buffer is 222 * always aligned to a 64 bit boundary. 223 * 224 * The size of the buffer is in bytes, but is it assumed that it is 225 * always ok to read a complete multiple of 64 bits at the end 226 * of the block in case the end is no aligned to a natural boundary. 227 * 228 * Return: the block number (bitmap buffer scope) that was found 229 */ 230 231 static u32 gfs2_bitfit(const u8 *buf, const unsigned int len, 232 u32 goal, u8 state) 233 { 234 u32 spoint = (goal << 1) & ((8*sizeof(u64)) - 1); 235 const __le64 *ptr = ((__le64 *)buf) + (goal >> 5); 236 const __le64 *end = (__le64 *)(buf + ALIGN(len, sizeof(u64))); 237 u64 tmp; 238 u64 mask = 0x5555555555555555ULL; 239 u32 bit; 240 241 /* Mask off bits we don't care about at the start of the search */ 242 mask <<= spoint; 243 tmp = gfs2_bit_search(ptr, mask, state); 244 ptr++; 245 while(tmp == 0 && ptr < end) { 246 tmp = gfs2_bit_search(ptr, 0x5555555555555555ULL, state); 247 ptr++; 248 } 249 /* Mask off any bits which are more than len bytes from the start */ 250 if (ptr == end && (len & (sizeof(u64) - 1))) 251 tmp &= (((u64)~0) >> (64 - 8*(len & (sizeof(u64) - 1)))); 252 /* Didn't find anything, so return */ 253 if (tmp == 0) 254 return BFITNOENT; 255 ptr--; 256 bit = __ffs64(tmp); 257 bit /= 2; /* two bits per entry in the bitmap */ 258 return (((const unsigned char *)ptr - buf) * GFS2_NBBY) + bit; 259 } 260 261 /** 262 * gfs2_rbm_from_block - Set the rbm based upon rgd and block number 263 * @rbm: The rbm with rgd already set correctly 264 * @block: The block number (filesystem relative) 265 * 266 * This sets the bi and offset members of an rbm based on a 267 * resource group and a filesystem relative block number. The 268 * resource group must be set in the rbm on entry, the bi and 269 * offset members will be set by this function. 270 * 271 * Returns: 0 on success, or an error code 272 */ 273 274 static int gfs2_rbm_from_block(struct gfs2_rbm *rbm, u64 block) 275 { 276 if (!rgrp_contains_block(rbm->rgd, block)) 277 return -E2BIG; 278 rbm->bii = 0; 279 rbm->offset = block - rbm->rgd->rd_data0; 280 /* Check if the block is within the first block */ 281 if (rbm->offset < rbm_bi(rbm)->bi_blocks) 282 return 0; 283 284 /* Adjust for the size diff between gfs2_meta_header and gfs2_rgrp */ 285 rbm->offset += (sizeof(struct gfs2_rgrp) - 286 sizeof(struct gfs2_meta_header)) * GFS2_NBBY; 287 rbm->bii = rbm->offset / rbm->rgd->rd_sbd->sd_blocks_per_bitmap; 288 rbm->offset -= rbm->bii * rbm->rgd->rd_sbd->sd_blocks_per_bitmap; 289 return 0; 290 } 291 292 /** 293 * gfs2_rbm_incr - increment an rbm structure 294 * @rbm: The rbm with rgd already set correctly 295 * 296 * This function takes an existing rbm structure and increments it to the next 297 * viable block offset. 298 * 299 * Returns: If incrementing the offset would cause the rbm to go past the 300 * end of the rgrp, true is returned, otherwise false. 301 * 302 */ 303 304 static bool gfs2_rbm_incr(struct gfs2_rbm *rbm) 305 { 306 if (rbm->offset + 1 < rbm_bi(rbm)->bi_blocks) { /* in the same bitmap */ 307 rbm->offset++; 308 return false; 309 } 310 if (rbm->bii == rbm->rgd->rd_length - 1) /* at the last bitmap */ 311 return true; 312 313 rbm->offset = 0; 314 rbm->bii++; 315 return false; 316 } 317 318 /** 319 * gfs2_unaligned_extlen - Look for free blocks which are not byte aligned 320 * @rbm: Position to search (value/result) 321 * @n_unaligned: Number of unaligned blocks to check 322 * @len: Decremented for each block found (terminate on zero) 323 * 324 * Returns: true if a non-free block is encountered 325 */ 326 327 static bool gfs2_unaligned_extlen(struct gfs2_rbm *rbm, u32 n_unaligned, u32 *len) 328 { 329 u32 n; 330 u8 res; 331 332 for (n = 0; n < n_unaligned; n++) { 333 res = gfs2_testbit(rbm, true); 334 if (res != GFS2_BLKST_FREE) 335 return true; 336 (*len)--; 337 if (*len == 0) 338 return true; 339 if (gfs2_rbm_incr(rbm)) 340 return true; 341 } 342 343 return false; 344 } 345 346 /** 347 * gfs2_free_extlen - Return extent length of free blocks 348 * @rrbm: Starting position 349 * @len: Max length to check 350 * 351 * Starting at the block specified by the rbm, see how many free blocks 352 * there are, not reading more than len blocks ahead. This can be done 353 * using memchr_inv when the blocks are byte aligned, but has to be done 354 * on a block by block basis in case of unaligned blocks. Also this 355 * function can cope with bitmap boundaries (although it must stop on 356 * a resource group boundary) 357 * 358 * Returns: Number of free blocks in the extent 359 */ 360 361 static u32 gfs2_free_extlen(const struct gfs2_rbm *rrbm, u32 len) 362 { 363 struct gfs2_rbm rbm = *rrbm; 364 u32 n_unaligned = rbm.offset & 3; 365 u32 size = len; 366 u32 bytes; 367 u32 chunk_size; 368 u8 *ptr, *start, *end; 369 u64 block; 370 struct gfs2_bitmap *bi; 371 372 if (n_unaligned && 373 gfs2_unaligned_extlen(&rbm, 4 - n_unaligned, &len)) 374 goto out; 375 376 n_unaligned = len & 3; 377 /* Start is now byte aligned */ 378 while (len > 3) { 379 bi = rbm_bi(&rbm); 380 start = bi->bi_bh->b_data; 381 if (bi->bi_clone) 382 start = bi->bi_clone; 383 start += bi->bi_offset; 384 end = start + bi->bi_bytes; 385 BUG_ON(rbm.offset & 3); 386 start += (rbm.offset / GFS2_NBBY); 387 bytes = min_t(u32, len / GFS2_NBBY, (end - start)); 388 ptr = memchr_inv(start, 0, bytes); 389 chunk_size = ((ptr == NULL) ? bytes : (ptr - start)); 390 chunk_size *= GFS2_NBBY; 391 BUG_ON(len < chunk_size); 392 len -= chunk_size; 393 block = gfs2_rbm_to_block(&rbm); 394 if (gfs2_rbm_from_block(&rbm, block + chunk_size)) { 395 n_unaligned = 0; 396 break; 397 } 398 if (ptr) { 399 n_unaligned = 3; 400 break; 401 } 402 n_unaligned = len & 3; 403 } 404 405 /* Deal with any bits left over at the end */ 406 if (n_unaligned) 407 gfs2_unaligned_extlen(&rbm, n_unaligned, &len); 408 out: 409 return size - len; 410 } 411 412 /** 413 * gfs2_bitcount - count the number of bits in a certain state 414 * @rgd: the resource group descriptor 415 * @buffer: the buffer that holds the bitmaps 416 * @buflen: the length (in bytes) of the buffer 417 * @state: the state of the block we're looking for 418 * 419 * Returns: The number of bits 420 */ 421 422 static u32 gfs2_bitcount(struct gfs2_rgrpd *rgd, const u8 *buffer, 423 unsigned int buflen, u8 state) 424 { 425 const u8 *byte = buffer; 426 const u8 *end = buffer + buflen; 427 const u8 state1 = state << 2; 428 const u8 state2 = state << 4; 429 const u8 state3 = state << 6; 430 u32 count = 0; 431 432 for (; byte < end; byte++) { 433 if (((*byte) & 0x03) == state) 434 count++; 435 if (((*byte) & 0x0C) == state1) 436 count++; 437 if (((*byte) & 0x30) == state2) 438 count++; 439 if (((*byte) & 0xC0) == state3) 440 count++; 441 } 442 443 return count; 444 } 445 446 /** 447 * gfs2_rgrp_verify - Verify that a resource group is consistent 448 * @rgd: the rgrp 449 * 450 */ 451 452 void gfs2_rgrp_verify(struct gfs2_rgrpd *rgd) 453 { 454 struct gfs2_sbd *sdp = rgd->rd_sbd; 455 struct gfs2_bitmap *bi = NULL; 456 u32 length = rgd->rd_length; 457 u32 count[4], tmp; 458 int buf, x; 459 460 memset(count, 0, 4 * sizeof(u32)); 461 462 /* Count # blocks in each of 4 possible allocation states */ 463 for (buf = 0; buf < length; buf++) { 464 bi = rgd->rd_bits + buf; 465 for (x = 0; x < 4; x++) 466 count[x] += gfs2_bitcount(rgd, 467 bi->bi_bh->b_data + 468 bi->bi_offset, 469 bi->bi_bytes, x); 470 } 471 472 if (count[0] != rgd->rd_free) { 473 if (gfs2_consist_rgrpd(rgd)) 474 fs_err(sdp, "free data mismatch: %u != %u\n", 475 count[0], rgd->rd_free); 476 return; 477 } 478 479 tmp = rgd->rd_data - rgd->rd_free - rgd->rd_dinodes; 480 if (count[1] != tmp) { 481 if (gfs2_consist_rgrpd(rgd)) 482 fs_err(sdp, "used data mismatch: %u != %u\n", 483 count[1], tmp); 484 return; 485 } 486 487 if (count[2] + count[3] != rgd->rd_dinodes) { 488 if (gfs2_consist_rgrpd(rgd)) 489 fs_err(sdp, "used metadata mismatch: %u != %u\n", 490 count[2] + count[3], rgd->rd_dinodes); 491 return; 492 } 493 } 494 495 /** 496 * gfs2_blk2rgrpd - Find resource group for a given data/meta block number 497 * @sdp: The GFS2 superblock 498 * @blk: The data block number 499 * @exact: True if this needs to be an exact match 500 * 501 * The @exact argument should be set to true by most callers. The exception 502 * is when we need to match blocks which are not represented by the rgrp 503 * bitmap, but which are part of the rgrp (i.e. padding blocks) which are 504 * there for alignment purposes. Another way of looking at it is that @exact 505 * matches only valid data/metadata blocks, but with @exact false, it will 506 * match any block within the extent of the rgrp. 507 * 508 * Returns: The resource group, or NULL if not found 509 */ 510 511 struct gfs2_rgrpd *gfs2_blk2rgrpd(struct gfs2_sbd *sdp, u64 blk, bool exact) 512 { 513 struct rb_node *n, *next; 514 struct gfs2_rgrpd *cur; 515 516 spin_lock(&sdp->sd_rindex_spin); 517 n = sdp->sd_rindex_tree.rb_node; 518 while (n) { 519 cur = rb_entry(n, struct gfs2_rgrpd, rd_node); 520 next = NULL; 521 if (blk < cur->rd_addr) 522 next = n->rb_left; 523 else if (blk >= cur->rd_data0 + cur->rd_data) 524 next = n->rb_right; 525 if (next == NULL) { 526 spin_unlock(&sdp->sd_rindex_spin); 527 if (exact) { 528 if (blk < cur->rd_addr) 529 return NULL; 530 if (blk >= cur->rd_data0 + cur->rd_data) 531 return NULL; 532 } 533 return cur; 534 } 535 n = next; 536 } 537 spin_unlock(&sdp->sd_rindex_spin); 538 539 return NULL; 540 } 541 542 /** 543 * gfs2_rgrpd_get_first - get the first Resource Group in the filesystem 544 * @sdp: The GFS2 superblock 545 * 546 * Returns: The first rgrp in the filesystem 547 */ 548 549 struct gfs2_rgrpd *gfs2_rgrpd_get_first(struct gfs2_sbd *sdp) 550 { 551 const struct rb_node *n; 552 struct gfs2_rgrpd *rgd; 553 554 spin_lock(&sdp->sd_rindex_spin); 555 n = rb_first(&sdp->sd_rindex_tree); 556 rgd = rb_entry(n, struct gfs2_rgrpd, rd_node); 557 spin_unlock(&sdp->sd_rindex_spin); 558 559 return rgd; 560 } 561 562 /** 563 * gfs2_rgrpd_get_next - get the next RG 564 * @rgd: the resource group descriptor 565 * 566 * Returns: The next rgrp 567 */ 568 569 struct gfs2_rgrpd *gfs2_rgrpd_get_next(struct gfs2_rgrpd *rgd) 570 { 571 struct gfs2_sbd *sdp = rgd->rd_sbd; 572 const struct rb_node *n; 573 574 spin_lock(&sdp->sd_rindex_spin); 575 n = rb_next(&rgd->rd_node); 576 if (n == NULL) 577 n = rb_first(&sdp->sd_rindex_tree); 578 579 if (unlikely(&rgd->rd_node == n)) { 580 spin_unlock(&sdp->sd_rindex_spin); 581 return NULL; 582 } 583 rgd = rb_entry(n, struct gfs2_rgrpd, rd_node); 584 spin_unlock(&sdp->sd_rindex_spin); 585 return rgd; 586 } 587 588 void check_and_update_goal(struct gfs2_inode *ip) 589 { 590 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 591 if (!ip->i_goal || gfs2_blk2rgrpd(sdp, ip->i_goal, 1) == NULL) 592 ip->i_goal = ip->i_no_addr; 593 } 594 595 void gfs2_free_clones(struct gfs2_rgrpd *rgd) 596 { 597 int x; 598 599 for (x = 0; x < rgd->rd_length; x++) { 600 struct gfs2_bitmap *bi = rgd->rd_bits + x; 601 kfree(bi->bi_clone); 602 bi->bi_clone = NULL; 603 } 604 } 605 606 /** 607 * gfs2_rsqa_alloc - make sure we have a reservation assigned to the inode 608 * plus a quota allocations data structure, if necessary 609 * @ip: the inode for this reservation 610 */ 611 int gfs2_rsqa_alloc(struct gfs2_inode *ip) 612 { 613 return gfs2_qa_alloc(ip); 614 } 615 616 static void dump_rs(struct seq_file *seq, const struct gfs2_blkreserv *rs) 617 { 618 struct gfs2_inode *ip = container_of(rs, struct gfs2_inode, i_res); 619 620 gfs2_print_dbg(seq, " B: n:%llu s:%llu b:%u f:%u\n", 621 (unsigned long long)ip->i_no_addr, 622 (unsigned long long)gfs2_rbm_to_block(&rs->rs_rbm), 623 rs->rs_rbm.offset, rs->rs_free); 624 } 625 626 /** 627 * __rs_deltree - remove a multi-block reservation from the rgd tree 628 * @rs: The reservation to remove 629 * 630 */ 631 static void __rs_deltree(struct gfs2_blkreserv *rs) 632 { 633 struct gfs2_rgrpd *rgd; 634 635 if (!gfs2_rs_active(rs)) 636 return; 637 638 rgd = rs->rs_rbm.rgd; 639 trace_gfs2_rs(rs, TRACE_RS_TREEDEL); 640 rb_erase(&rs->rs_node, &rgd->rd_rstree); 641 RB_CLEAR_NODE(&rs->rs_node); 642 643 if (rs->rs_free) { 644 u64 last_block = gfs2_rbm_to_block(&rs->rs_rbm) + 645 rs->rs_free - 1; 646 struct gfs2_rbm last_rbm = { .rgd = rs->rs_rbm.rgd, }; 647 struct gfs2_bitmap *start, *last; 648 649 /* return reserved blocks to the rgrp */ 650 BUG_ON(rs->rs_rbm.rgd->rd_reserved < rs->rs_free); 651 rs->rs_rbm.rgd->rd_reserved -= rs->rs_free; 652 /* The rgrp extent failure point is likely not to increase; 653 it will only do so if the freed blocks are somehow 654 contiguous with a span of free blocks that follows. Still, 655 it will force the number to be recalculated later. */ 656 rgd->rd_extfail_pt += rs->rs_free; 657 rs->rs_free = 0; 658 if (gfs2_rbm_from_block(&last_rbm, last_block)) 659 return; 660 start = rbm_bi(&rs->rs_rbm); 661 last = rbm_bi(&last_rbm); 662 do 663 clear_bit(GBF_FULL, &start->bi_flags); 664 while (start++ != last); 665 } 666 } 667 668 /** 669 * gfs2_rs_deltree - remove a multi-block reservation from the rgd tree 670 * @rs: The reservation to remove 671 * 672 */ 673 void gfs2_rs_deltree(struct gfs2_blkreserv *rs) 674 { 675 struct gfs2_rgrpd *rgd; 676 677 rgd = rs->rs_rbm.rgd; 678 if (rgd) { 679 spin_lock(&rgd->rd_rsspin); 680 __rs_deltree(rs); 681 BUG_ON(rs->rs_free); 682 spin_unlock(&rgd->rd_rsspin); 683 } 684 } 685 686 /** 687 * gfs2_rsqa_delete - delete a multi-block reservation and quota allocation 688 * @ip: The inode for this reservation 689 * @wcount: The inode's write count, or NULL 690 * 691 */ 692 void gfs2_rsqa_delete(struct gfs2_inode *ip, atomic_t *wcount) 693 { 694 down_write(&ip->i_rw_mutex); 695 if ((wcount == NULL) || (atomic_read(wcount) <= 1)) 696 gfs2_rs_deltree(&ip->i_res); 697 up_write(&ip->i_rw_mutex); 698 gfs2_qa_delete(ip, wcount); 699 } 700 701 /** 702 * return_all_reservations - return all reserved blocks back to the rgrp. 703 * @rgd: the rgrp that needs its space back 704 * 705 * We previously reserved a bunch of blocks for allocation. Now we need to 706 * give them back. This leave the reservation structures in tact, but removes 707 * all of their corresponding "no-fly zones". 708 */ 709 static void return_all_reservations(struct gfs2_rgrpd *rgd) 710 { 711 struct rb_node *n; 712 struct gfs2_blkreserv *rs; 713 714 spin_lock(&rgd->rd_rsspin); 715 while ((n = rb_first(&rgd->rd_rstree))) { 716 rs = rb_entry(n, struct gfs2_blkreserv, rs_node); 717 __rs_deltree(rs); 718 } 719 spin_unlock(&rgd->rd_rsspin); 720 } 721 722 void gfs2_clear_rgrpd(struct gfs2_sbd *sdp) 723 { 724 struct rb_node *n; 725 struct gfs2_rgrpd *rgd; 726 struct gfs2_glock *gl; 727 728 while ((n = rb_first(&sdp->sd_rindex_tree))) { 729 rgd = rb_entry(n, struct gfs2_rgrpd, rd_node); 730 gl = rgd->rd_gl; 731 732 rb_erase(n, &sdp->sd_rindex_tree); 733 734 if (gl) { 735 glock_clear_object(gl, rgd); 736 gfs2_glock_put(gl); 737 } 738 739 gfs2_free_clones(rgd); 740 kfree(rgd->rd_bits); 741 rgd->rd_bits = NULL; 742 return_all_reservations(rgd); 743 kmem_cache_free(gfs2_rgrpd_cachep, rgd); 744 } 745 } 746 747 static void gfs2_rindex_print(const struct gfs2_rgrpd *rgd) 748 { 749 struct gfs2_sbd *sdp = rgd->rd_sbd; 750 751 fs_info(sdp, "ri_addr = %llu\n", (unsigned long long)rgd->rd_addr); 752 fs_info(sdp, "ri_length = %u\n", rgd->rd_length); 753 fs_info(sdp, "ri_data0 = %llu\n", (unsigned long long)rgd->rd_data0); 754 fs_info(sdp, "ri_data = %u\n", rgd->rd_data); 755 fs_info(sdp, "ri_bitbytes = %u\n", rgd->rd_bitbytes); 756 } 757 758 /** 759 * gfs2_compute_bitstructs - Compute the bitmap sizes 760 * @rgd: The resource group descriptor 761 * 762 * Calculates bitmap descriptors, one for each block that contains bitmap data 763 * 764 * Returns: errno 765 */ 766 767 static int compute_bitstructs(struct gfs2_rgrpd *rgd) 768 { 769 struct gfs2_sbd *sdp = rgd->rd_sbd; 770 struct gfs2_bitmap *bi; 771 u32 length = rgd->rd_length; /* # blocks in hdr & bitmap */ 772 u32 bytes_left, bytes; 773 int x; 774 775 if (!length) 776 return -EINVAL; 777 778 rgd->rd_bits = kcalloc(length, sizeof(struct gfs2_bitmap), GFP_NOFS); 779 if (!rgd->rd_bits) 780 return -ENOMEM; 781 782 bytes_left = rgd->rd_bitbytes; 783 784 for (x = 0; x < length; x++) { 785 bi = rgd->rd_bits + x; 786 787 bi->bi_flags = 0; 788 /* small rgrp; bitmap stored completely in header block */ 789 if (length == 1) { 790 bytes = bytes_left; 791 bi->bi_offset = sizeof(struct gfs2_rgrp); 792 bi->bi_start = 0; 793 bi->bi_bytes = bytes; 794 bi->bi_blocks = bytes * GFS2_NBBY; 795 /* header block */ 796 } else if (x == 0) { 797 bytes = sdp->sd_sb.sb_bsize - sizeof(struct gfs2_rgrp); 798 bi->bi_offset = sizeof(struct gfs2_rgrp); 799 bi->bi_start = 0; 800 bi->bi_bytes = bytes; 801 bi->bi_blocks = bytes * GFS2_NBBY; 802 /* last block */ 803 } else if (x + 1 == length) { 804 bytes = bytes_left; 805 bi->bi_offset = sizeof(struct gfs2_meta_header); 806 bi->bi_start = rgd->rd_bitbytes - bytes_left; 807 bi->bi_bytes = bytes; 808 bi->bi_blocks = bytes * GFS2_NBBY; 809 /* other blocks */ 810 } else { 811 bytes = sdp->sd_sb.sb_bsize - 812 sizeof(struct gfs2_meta_header); 813 bi->bi_offset = sizeof(struct gfs2_meta_header); 814 bi->bi_start = rgd->rd_bitbytes - bytes_left; 815 bi->bi_bytes = bytes; 816 bi->bi_blocks = bytes * GFS2_NBBY; 817 } 818 819 bytes_left -= bytes; 820 } 821 822 if (bytes_left) { 823 gfs2_consist_rgrpd(rgd); 824 return -EIO; 825 } 826 bi = rgd->rd_bits + (length - 1); 827 if ((bi->bi_start + bi->bi_bytes) * GFS2_NBBY != rgd->rd_data) { 828 if (gfs2_consist_rgrpd(rgd)) { 829 gfs2_rindex_print(rgd); 830 fs_err(sdp, "start=%u len=%u offset=%u\n", 831 bi->bi_start, bi->bi_bytes, bi->bi_offset); 832 } 833 return -EIO; 834 } 835 836 return 0; 837 } 838 839 /** 840 * gfs2_ri_total - Total up the file system space, according to the rindex. 841 * @sdp: the filesystem 842 * 843 */ 844 u64 gfs2_ri_total(struct gfs2_sbd *sdp) 845 { 846 u64 total_data = 0; 847 struct inode *inode = sdp->sd_rindex; 848 struct gfs2_inode *ip = GFS2_I(inode); 849 char buf[sizeof(struct gfs2_rindex)]; 850 int error, rgrps; 851 852 for (rgrps = 0;; rgrps++) { 853 loff_t pos = rgrps * sizeof(struct gfs2_rindex); 854 855 if (pos + sizeof(struct gfs2_rindex) > i_size_read(inode)) 856 break; 857 error = gfs2_internal_read(ip, buf, &pos, 858 sizeof(struct gfs2_rindex)); 859 if (error != sizeof(struct gfs2_rindex)) 860 break; 861 total_data += be32_to_cpu(((struct gfs2_rindex *)buf)->ri_data); 862 } 863 return total_data; 864 } 865 866 static int rgd_insert(struct gfs2_rgrpd *rgd) 867 { 868 struct gfs2_sbd *sdp = rgd->rd_sbd; 869 struct rb_node **newn = &sdp->sd_rindex_tree.rb_node, *parent = NULL; 870 871 /* Figure out where to put new node */ 872 while (*newn) { 873 struct gfs2_rgrpd *cur = rb_entry(*newn, struct gfs2_rgrpd, 874 rd_node); 875 876 parent = *newn; 877 if (rgd->rd_addr < cur->rd_addr) 878 newn = &((*newn)->rb_left); 879 else if (rgd->rd_addr > cur->rd_addr) 880 newn = &((*newn)->rb_right); 881 else 882 return -EEXIST; 883 } 884 885 rb_link_node(&rgd->rd_node, parent, newn); 886 rb_insert_color(&rgd->rd_node, &sdp->sd_rindex_tree); 887 sdp->sd_rgrps++; 888 return 0; 889 } 890 891 /** 892 * read_rindex_entry - Pull in a new resource index entry from the disk 893 * @ip: Pointer to the rindex inode 894 * 895 * Returns: 0 on success, > 0 on EOF, error code otherwise 896 */ 897 898 static int read_rindex_entry(struct gfs2_inode *ip) 899 { 900 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 901 const unsigned bsize = sdp->sd_sb.sb_bsize; 902 loff_t pos = sdp->sd_rgrps * sizeof(struct gfs2_rindex); 903 struct gfs2_rindex buf; 904 int error; 905 struct gfs2_rgrpd *rgd; 906 907 if (pos >= i_size_read(&ip->i_inode)) 908 return 1; 909 910 error = gfs2_internal_read(ip, (char *)&buf, &pos, 911 sizeof(struct gfs2_rindex)); 912 913 if (error != sizeof(struct gfs2_rindex)) 914 return (error == 0) ? 1 : error; 915 916 rgd = kmem_cache_zalloc(gfs2_rgrpd_cachep, GFP_NOFS); 917 error = -ENOMEM; 918 if (!rgd) 919 return error; 920 921 rgd->rd_sbd = sdp; 922 rgd->rd_addr = be64_to_cpu(buf.ri_addr); 923 rgd->rd_length = be32_to_cpu(buf.ri_length); 924 rgd->rd_data0 = be64_to_cpu(buf.ri_data0); 925 rgd->rd_data = be32_to_cpu(buf.ri_data); 926 rgd->rd_bitbytes = be32_to_cpu(buf.ri_bitbytes); 927 spin_lock_init(&rgd->rd_rsspin); 928 929 error = compute_bitstructs(rgd); 930 if (error) 931 goto fail; 932 933 error = gfs2_glock_get(sdp, rgd->rd_addr, 934 &gfs2_rgrp_glops, CREATE, &rgd->rd_gl); 935 if (error) 936 goto fail; 937 938 rgd->rd_rgl = (struct gfs2_rgrp_lvb *)rgd->rd_gl->gl_lksb.sb_lvbptr; 939 rgd->rd_flags &= ~(GFS2_RDF_UPTODATE | GFS2_RDF_PREFERRED); 940 if (rgd->rd_data > sdp->sd_max_rg_data) 941 sdp->sd_max_rg_data = rgd->rd_data; 942 spin_lock(&sdp->sd_rindex_spin); 943 error = rgd_insert(rgd); 944 spin_unlock(&sdp->sd_rindex_spin); 945 if (!error) { 946 glock_set_object(rgd->rd_gl, rgd); 947 rgd->rd_gl->gl_vm.start = (rgd->rd_addr * bsize) & PAGE_MASK; 948 rgd->rd_gl->gl_vm.end = PAGE_ALIGN((rgd->rd_addr + 949 rgd->rd_length) * bsize) - 1; 950 return 0; 951 } 952 953 error = 0; /* someone else read in the rgrp; free it and ignore it */ 954 gfs2_glock_put(rgd->rd_gl); 955 956 fail: 957 kfree(rgd->rd_bits); 958 rgd->rd_bits = NULL; 959 kmem_cache_free(gfs2_rgrpd_cachep, rgd); 960 return error; 961 } 962 963 /** 964 * set_rgrp_preferences - Run all the rgrps, selecting some we prefer to use 965 * @sdp: the GFS2 superblock 966 * 967 * The purpose of this function is to select a subset of the resource groups 968 * and mark them as PREFERRED. We do it in such a way that each node prefers 969 * to use a unique set of rgrps to minimize glock contention. 970 */ 971 static void set_rgrp_preferences(struct gfs2_sbd *sdp) 972 { 973 struct gfs2_rgrpd *rgd, *first; 974 int i; 975 976 /* Skip an initial number of rgrps, based on this node's journal ID. 977 That should start each node out on its own set. */ 978 rgd = gfs2_rgrpd_get_first(sdp); 979 for (i = 0; i < sdp->sd_lockstruct.ls_jid; i++) 980 rgd = gfs2_rgrpd_get_next(rgd); 981 first = rgd; 982 983 do { 984 rgd->rd_flags |= GFS2_RDF_PREFERRED; 985 for (i = 0; i < sdp->sd_journals; i++) { 986 rgd = gfs2_rgrpd_get_next(rgd); 987 if (!rgd || rgd == first) 988 break; 989 } 990 } while (rgd && rgd != first); 991 } 992 993 /** 994 * gfs2_ri_update - Pull in a new resource index from the disk 995 * @ip: pointer to the rindex inode 996 * 997 * Returns: 0 on successful update, error code otherwise 998 */ 999 1000 static int gfs2_ri_update(struct gfs2_inode *ip) 1001 { 1002 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 1003 int error; 1004 1005 do { 1006 error = read_rindex_entry(ip); 1007 } while (error == 0); 1008 1009 if (error < 0) 1010 return error; 1011 1012 set_rgrp_preferences(sdp); 1013 1014 sdp->sd_rindex_uptodate = 1; 1015 return 0; 1016 } 1017 1018 /** 1019 * gfs2_rindex_update - Update the rindex if required 1020 * @sdp: The GFS2 superblock 1021 * 1022 * We grab a lock on the rindex inode to make sure that it doesn't 1023 * change whilst we are performing an operation. We keep this lock 1024 * for quite long periods of time compared to other locks. This 1025 * doesn't matter, since it is shared and it is very, very rarely 1026 * accessed in the exclusive mode (i.e. only when expanding the filesystem). 1027 * 1028 * This makes sure that we're using the latest copy of the resource index 1029 * special file, which might have been updated if someone expanded the 1030 * filesystem (via gfs2_grow utility), which adds new resource groups. 1031 * 1032 * Returns: 0 on succeess, error code otherwise 1033 */ 1034 1035 int gfs2_rindex_update(struct gfs2_sbd *sdp) 1036 { 1037 struct gfs2_inode *ip = GFS2_I(sdp->sd_rindex); 1038 struct gfs2_glock *gl = ip->i_gl; 1039 struct gfs2_holder ri_gh; 1040 int error = 0; 1041 int unlock_required = 0; 1042 1043 /* Read new copy from disk if we don't have the latest */ 1044 if (!sdp->sd_rindex_uptodate) { 1045 if (!gfs2_glock_is_locked_by_me(gl)) { 1046 error = gfs2_glock_nq_init(gl, LM_ST_SHARED, 0, &ri_gh); 1047 if (error) 1048 return error; 1049 unlock_required = 1; 1050 } 1051 if (!sdp->sd_rindex_uptodate) 1052 error = gfs2_ri_update(ip); 1053 if (unlock_required) 1054 gfs2_glock_dq_uninit(&ri_gh); 1055 } 1056 1057 return error; 1058 } 1059 1060 static void gfs2_rgrp_in(struct gfs2_rgrpd *rgd, const void *buf) 1061 { 1062 const struct gfs2_rgrp *str = buf; 1063 u32 rg_flags; 1064 1065 rg_flags = be32_to_cpu(str->rg_flags); 1066 rg_flags &= ~GFS2_RDF_MASK; 1067 rgd->rd_flags &= GFS2_RDF_MASK; 1068 rgd->rd_flags |= rg_flags; 1069 rgd->rd_free = be32_to_cpu(str->rg_free); 1070 rgd->rd_dinodes = be32_to_cpu(str->rg_dinodes); 1071 rgd->rd_igeneration = be64_to_cpu(str->rg_igeneration); 1072 /* rd_data0, rd_data and rd_bitbytes already set from rindex */ 1073 } 1074 1075 static void gfs2_rgrp_ondisk2lvb(struct gfs2_rgrp_lvb *rgl, const void *buf) 1076 { 1077 const struct gfs2_rgrp *str = buf; 1078 1079 rgl->rl_magic = cpu_to_be32(GFS2_MAGIC); 1080 rgl->rl_flags = str->rg_flags; 1081 rgl->rl_free = str->rg_free; 1082 rgl->rl_dinodes = str->rg_dinodes; 1083 rgl->rl_igeneration = str->rg_igeneration; 1084 rgl->__pad = 0UL; 1085 } 1086 1087 static void gfs2_rgrp_out(struct gfs2_rgrpd *rgd, void *buf) 1088 { 1089 struct gfs2_rgrpd *next = gfs2_rgrpd_get_next(rgd); 1090 struct gfs2_rgrp *str = buf; 1091 u32 crc; 1092 1093 str->rg_flags = cpu_to_be32(rgd->rd_flags & ~GFS2_RDF_MASK); 1094 str->rg_free = cpu_to_be32(rgd->rd_free); 1095 str->rg_dinodes = cpu_to_be32(rgd->rd_dinodes); 1096 if (next == NULL) 1097 str->rg_skip = 0; 1098 else if (next->rd_addr > rgd->rd_addr) 1099 str->rg_skip = cpu_to_be32(next->rd_addr - rgd->rd_addr); 1100 str->rg_igeneration = cpu_to_be64(rgd->rd_igeneration); 1101 str->rg_data0 = cpu_to_be64(rgd->rd_data0); 1102 str->rg_data = cpu_to_be32(rgd->rd_data); 1103 str->rg_bitbytes = cpu_to_be32(rgd->rd_bitbytes); 1104 str->rg_crc = 0; 1105 crc = gfs2_disk_hash(buf, sizeof(struct gfs2_rgrp)); 1106 str->rg_crc = cpu_to_be32(crc); 1107 1108 memset(&str->rg_reserved, 0, sizeof(str->rg_reserved)); 1109 gfs2_rgrp_ondisk2lvb(rgd->rd_rgl, buf); 1110 } 1111 1112 static int gfs2_rgrp_lvb_valid(struct gfs2_rgrpd *rgd) 1113 { 1114 struct gfs2_rgrp_lvb *rgl = rgd->rd_rgl; 1115 struct gfs2_rgrp *str = (struct gfs2_rgrp *)rgd->rd_bits[0].bi_bh->b_data; 1116 int valid = 1; 1117 1118 if (rgl->rl_flags != str->rg_flags) { 1119 printk(KERN_WARNING "GFS2: rgd: %llu lvb flag mismatch %u/%u", 1120 (unsigned long long)rgd->rd_addr, 1121 be32_to_cpu(rgl->rl_flags), be32_to_cpu(str->rg_flags)); 1122 valid = 0; 1123 } 1124 if (rgl->rl_free != str->rg_free) { 1125 printk(KERN_WARNING "GFS2: rgd: %llu lvb free mismatch %u/%u", 1126 (unsigned long long)rgd->rd_addr, 1127 be32_to_cpu(rgl->rl_free), be32_to_cpu(str->rg_free)); 1128 valid = 0; 1129 } 1130 if (rgl->rl_dinodes != str->rg_dinodes) { 1131 printk(KERN_WARNING "GFS2: rgd: %llu lvb dinode mismatch %u/%u", 1132 (unsigned long long)rgd->rd_addr, 1133 be32_to_cpu(rgl->rl_dinodes), 1134 be32_to_cpu(str->rg_dinodes)); 1135 valid = 0; 1136 } 1137 if (rgl->rl_igeneration != str->rg_igeneration) { 1138 printk(KERN_WARNING "GFS2: rgd: %llu lvb igen mismatch " 1139 "%llu/%llu", (unsigned long long)rgd->rd_addr, 1140 (unsigned long long)be64_to_cpu(rgl->rl_igeneration), 1141 (unsigned long long)be64_to_cpu(str->rg_igeneration)); 1142 valid = 0; 1143 } 1144 return valid; 1145 } 1146 1147 static u32 count_unlinked(struct gfs2_rgrpd *rgd) 1148 { 1149 struct gfs2_bitmap *bi; 1150 const u32 length = rgd->rd_length; 1151 const u8 *buffer = NULL; 1152 u32 i, goal, count = 0; 1153 1154 for (i = 0, bi = rgd->rd_bits; i < length; i++, bi++) { 1155 goal = 0; 1156 buffer = bi->bi_bh->b_data + bi->bi_offset; 1157 WARN_ON(!buffer_uptodate(bi->bi_bh)); 1158 while (goal < bi->bi_blocks) { 1159 goal = gfs2_bitfit(buffer, bi->bi_bytes, goal, 1160 GFS2_BLKST_UNLINKED); 1161 if (goal == BFITNOENT) 1162 break; 1163 count++; 1164 goal++; 1165 } 1166 } 1167 1168 return count; 1169 } 1170 1171 1172 /** 1173 * gfs2_rgrp_bh_get - Read in a RG's header and bitmaps 1174 * @rgd: the struct gfs2_rgrpd describing the RG to read in 1175 * 1176 * Read in all of a Resource Group's header and bitmap blocks. 1177 * Caller must eventually call gfs2_rgrp_relse() to free the bitmaps. 1178 * 1179 * Returns: errno 1180 */ 1181 1182 static int gfs2_rgrp_bh_get(struct gfs2_rgrpd *rgd) 1183 { 1184 struct gfs2_sbd *sdp = rgd->rd_sbd; 1185 struct gfs2_glock *gl = rgd->rd_gl; 1186 unsigned int length = rgd->rd_length; 1187 struct gfs2_bitmap *bi; 1188 unsigned int x, y; 1189 int error; 1190 1191 if (rgd->rd_bits[0].bi_bh != NULL) 1192 return 0; 1193 1194 for (x = 0; x < length; x++) { 1195 bi = rgd->rd_bits + x; 1196 error = gfs2_meta_read(gl, rgd->rd_addr + x, 0, 0, &bi->bi_bh); 1197 if (error) 1198 goto fail; 1199 } 1200 1201 for (y = length; y--;) { 1202 bi = rgd->rd_bits + y; 1203 error = gfs2_meta_wait(sdp, bi->bi_bh); 1204 if (error) 1205 goto fail; 1206 if (gfs2_metatype_check(sdp, bi->bi_bh, y ? GFS2_METATYPE_RB : 1207 GFS2_METATYPE_RG)) { 1208 error = -EIO; 1209 goto fail; 1210 } 1211 } 1212 1213 if (!(rgd->rd_flags & GFS2_RDF_UPTODATE)) { 1214 for (x = 0; x < length; x++) 1215 clear_bit(GBF_FULL, &rgd->rd_bits[x].bi_flags); 1216 gfs2_rgrp_in(rgd, (rgd->rd_bits[0].bi_bh)->b_data); 1217 rgd->rd_flags |= (GFS2_RDF_UPTODATE | GFS2_RDF_CHECK); 1218 rgd->rd_free_clone = rgd->rd_free; 1219 /* max out the rgrp allocation failure point */ 1220 rgd->rd_extfail_pt = rgd->rd_free; 1221 } 1222 if (cpu_to_be32(GFS2_MAGIC) != rgd->rd_rgl->rl_magic) { 1223 rgd->rd_rgl->rl_unlinked = cpu_to_be32(count_unlinked(rgd)); 1224 gfs2_rgrp_ondisk2lvb(rgd->rd_rgl, 1225 rgd->rd_bits[0].bi_bh->b_data); 1226 } 1227 else if (sdp->sd_args.ar_rgrplvb) { 1228 if (!gfs2_rgrp_lvb_valid(rgd)){ 1229 gfs2_consist_rgrpd(rgd); 1230 error = -EIO; 1231 goto fail; 1232 } 1233 if (rgd->rd_rgl->rl_unlinked == 0) 1234 rgd->rd_flags &= ~GFS2_RDF_CHECK; 1235 } 1236 return 0; 1237 1238 fail: 1239 while (x--) { 1240 bi = rgd->rd_bits + x; 1241 brelse(bi->bi_bh); 1242 bi->bi_bh = NULL; 1243 gfs2_assert_warn(sdp, !bi->bi_clone); 1244 } 1245 1246 return error; 1247 } 1248 1249 static int update_rgrp_lvb(struct gfs2_rgrpd *rgd) 1250 { 1251 u32 rl_flags; 1252 1253 if (rgd->rd_flags & GFS2_RDF_UPTODATE) 1254 return 0; 1255 1256 if (cpu_to_be32(GFS2_MAGIC) != rgd->rd_rgl->rl_magic) 1257 return gfs2_rgrp_bh_get(rgd); 1258 1259 rl_flags = be32_to_cpu(rgd->rd_rgl->rl_flags); 1260 rl_flags &= ~GFS2_RDF_MASK; 1261 rgd->rd_flags &= GFS2_RDF_MASK; 1262 rgd->rd_flags |= (rl_flags | GFS2_RDF_CHECK); 1263 if (rgd->rd_rgl->rl_unlinked == 0) 1264 rgd->rd_flags &= ~GFS2_RDF_CHECK; 1265 rgd->rd_free = be32_to_cpu(rgd->rd_rgl->rl_free); 1266 rgd->rd_free_clone = rgd->rd_free; 1267 rgd->rd_dinodes = be32_to_cpu(rgd->rd_rgl->rl_dinodes); 1268 rgd->rd_igeneration = be64_to_cpu(rgd->rd_rgl->rl_igeneration); 1269 return 0; 1270 } 1271 1272 int gfs2_rgrp_go_lock(struct gfs2_holder *gh) 1273 { 1274 struct gfs2_rgrpd *rgd = gh->gh_gl->gl_object; 1275 struct gfs2_sbd *sdp = rgd->rd_sbd; 1276 1277 if (gh->gh_flags & GL_SKIP && sdp->sd_args.ar_rgrplvb) 1278 return 0; 1279 return gfs2_rgrp_bh_get(rgd); 1280 } 1281 1282 /** 1283 * gfs2_rgrp_brelse - Release RG bitmaps read in with gfs2_rgrp_bh_get() 1284 * @rgd: The resource group 1285 * 1286 */ 1287 1288 void gfs2_rgrp_brelse(struct gfs2_rgrpd *rgd) 1289 { 1290 int x, length = rgd->rd_length; 1291 1292 for (x = 0; x < length; x++) { 1293 struct gfs2_bitmap *bi = rgd->rd_bits + x; 1294 if (bi->bi_bh) { 1295 brelse(bi->bi_bh); 1296 bi->bi_bh = NULL; 1297 } 1298 } 1299 1300 } 1301 1302 /** 1303 * gfs2_rgrp_go_unlock - Unlock a rgrp glock 1304 * @gh: The glock holder for the resource group 1305 * 1306 */ 1307 1308 void gfs2_rgrp_go_unlock(struct gfs2_holder *gh) 1309 { 1310 struct gfs2_rgrpd *rgd = gh->gh_gl->gl_object; 1311 int demote_requested = test_bit(GLF_DEMOTE, &gh->gh_gl->gl_flags) | 1312 test_bit(GLF_PENDING_DEMOTE, &gh->gh_gl->gl_flags); 1313 1314 if (rgd && demote_requested) 1315 gfs2_rgrp_brelse(rgd); 1316 } 1317 1318 int gfs2_rgrp_send_discards(struct gfs2_sbd *sdp, u64 offset, 1319 struct buffer_head *bh, 1320 const struct gfs2_bitmap *bi, unsigned minlen, u64 *ptrimmed) 1321 { 1322 struct super_block *sb = sdp->sd_vfs; 1323 u64 blk; 1324 sector_t start = 0; 1325 sector_t nr_blks = 0; 1326 int rv; 1327 unsigned int x; 1328 u32 trimmed = 0; 1329 u8 diff; 1330 1331 for (x = 0; x < bi->bi_bytes; x++) { 1332 const u8 *clone = bi->bi_clone ? bi->bi_clone : bi->bi_bh->b_data; 1333 clone += bi->bi_offset; 1334 clone += x; 1335 if (bh) { 1336 const u8 *orig = bh->b_data + bi->bi_offset + x; 1337 diff = ~(*orig | (*orig >> 1)) & (*clone | (*clone >> 1)); 1338 } else { 1339 diff = ~(*clone | (*clone >> 1)); 1340 } 1341 diff &= 0x55; 1342 if (diff == 0) 1343 continue; 1344 blk = offset + ((bi->bi_start + x) * GFS2_NBBY); 1345 while(diff) { 1346 if (diff & 1) { 1347 if (nr_blks == 0) 1348 goto start_new_extent; 1349 if ((start + nr_blks) != blk) { 1350 if (nr_blks >= minlen) { 1351 rv = sb_issue_discard(sb, 1352 start, nr_blks, 1353 GFP_NOFS, 0); 1354 if (rv) 1355 goto fail; 1356 trimmed += nr_blks; 1357 } 1358 nr_blks = 0; 1359 start_new_extent: 1360 start = blk; 1361 } 1362 nr_blks++; 1363 } 1364 diff >>= 2; 1365 blk++; 1366 } 1367 } 1368 if (nr_blks >= minlen) { 1369 rv = sb_issue_discard(sb, start, nr_blks, GFP_NOFS, 0); 1370 if (rv) 1371 goto fail; 1372 trimmed += nr_blks; 1373 } 1374 if (ptrimmed) 1375 *ptrimmed = trimmed; 1376 return 0; 1377 1378 fail: 1379 if (sdp->sd_args.ar_discard) 1380 fs_warn(sdp, "error %d on discard request, turning discards off for this filesystem\n", rv); 1381 sdp->sd_args.ar_discard = 0; 1382 return -EIO; 1383 } 1384 1385 /** 1386 * gfs2_fitrim - Generate discard requests for unused bits of the filesystem 1387 * @filp: Any file on the filesystem 1388 * @argp: Pointer to the arguments (also used to pass result) 1389 * 1390 * Returns: 0 on success, otherwise error code 1391 */ 1392 1393 int gfs2_fitrim(struct file *filp, void __user *argp) 1394 { 1395 struct inode *inode = file_inode(filp); 1396 struct gfs2_sbd *sdp = GFS2_SB(inode); 1397 struct request_queue *q = bdev_get_queue(sdp->sd_vfs->s_bdev); 1398 struct buffer_head *bh; 1399 struct gfs2_rgrpd *rgd; 1400 struct gfs2_rgrpd *rgd_end; 1401 struct gfs2_holder gh; 1402 struct fstrim_range r; 1403 int ret = 0; 1404 u64 amt; 1405 u64 trimmed = 0; 1406 u64 start, end, minlen; 1407 unsigned int x; 1408 unsigned bs_shift = sdp->sd_sb.sb_bsize_shift; 1409 1410 if (!capable(CAP_SYS_ADMIN)) 1411 return -EPERM; 1412 1413 if (!blk_queue_discard(q)) 1414 return -EOPNOTSUPP; 1415 1416 if (copy_from_user(&r, argp, sizeof(r))) 1417 return -EFAULT; 1418 1419 ret = gfs2_rindex_update(sdp); 1420 if (ret) 1421 return ret; 1422 1423 start = r.start >> bs_shift; 1424 end = start + (r.len >> bs_shift); 1425 minlen = max_t(u64, r.minlen, 1426 q->limits.discard_granularity) >> bs_shift; 1427 1428 if (end <= start || minlen > sdp->sd_max_rg_data) 1429 return -EINVAL; 1430 1431 rgd = gfs2_blk2rgrpd(sdp, start, 0); 1432 rgd_end = gfs2_blk2rgrpd(sdp, end, 0); 1433 1434 if ((gfs2_rgrpd_get_first(sdp) == gfs2_rgrpd_get_next(rgd_end)) 1435 && (start > rgd_end->rd_data0 + rgd_end->rd_data)) 1436 return -EINVAL; /* start is beyond the end of the fs */ 1437 1438 while (1) { 1439 1440 ret = gfs2_glock_nq_init(rgd->rd_gl, LM_ST_EXCLUSIVE, 0, &gh); 1441 if (ret) 1442 goto out; 1443 1444 if (!(rgd->rd_flags & GFS2_RGF_TRIMMED)) { 1445 /* Trim each bitmap in the rgrp */ 1446 for (x = 0; x < rgd->rd_length; x++) { 1447 struct gfs2_bitmap *bi = rgd->rd_bits + x; 1448 ret = gfs2_rgrp_send_discards(sdp, 1449 rgd->rd_data0, NULL, bi, minlen, 1450 &amt); 1451 if (ret) { 1452 gfs2_glock_dq_uninit(&gh); 1453 goto out; 1454 } 1455 trimmed += amt; 1456 } 1457 1458 /* Mark rgrp as having been trimmed */ 1459 ret = gfs2_trans_begin(sdp, RES_RG_HDR, 0); 1460 if (ret == 0) { 1461 bh = rgd->rd_bits[0].bi_bh; 1462 rgd->rd_flags |= GFS2_RGF_TRIMMED; 1463 gfs2_trans_add_meta(rgd->rd_gl, bh); 1464 gfs2_rgrp_out(rgd, bh->b_data); 1465 gfs2_trans_end(sdp); 1466 } 1467 } 1468 gfs2_glock_dq_uninit(&gh); 1469 1470 if (rgd == rgd_end) 1471 break; 1472 1473 rgd = gfs2_rgrpd_get_next(rgd); 1474 } 1475 1476 out: 1477 r.len = trimmed << bs_shift; 1478 if (copy_to_user(argp, &r, sizeof(r))) 1479 return -EFAULT; 1480 1481 return ret; 1482 } 1483 1484 /** 1485 * rs_insert - insert a new multi-block reservation into the rgrp's rb_tree 1486 * @ip: the inode structure 1487 * 1488 */ 1489 static void rs_insert(struct gfs2_inode *ip) 1490 { 1491 struct rb_node **newn, *parent = NULL; 1492 int rc; 1493 struct gfs2_blkreserv *rs = &ip->i_res; 1494 struct gfs2_rgrpd *rgd = rs->rs_rbm.rgd; 1495 u64 fsblock = gfs2_rbm_to_block(&rs->rs_rbm); 1496 1497 BUG_ON(gfs2_rs_active(rs)); 1498 1499 spin_lock(&rgd->rd_rsspin); 1500 newn = &rgd->rd_rstree.rb_node; 1501 while (*newn) { 1502 struct gfs2_blkreserv *cur = 1503 rb_entry(*newn, struct gfs2_blkreserv, rs_node); 1504 1505 parent = *newn; 1506 rc = rs_cmp(fsblock, rs->rs_free, cur); 1507 if (rc > 0) 1508 newn = &((*newn)->rb_right); 1509 else if (rc < 0) 1510 newn = &((*newn)->rb_left); 1511 else { 1512 spin_unlock(&rgd->rd_rsspin); 1513 WARN_ON(1); 1514 return; 1515 } 1516 } 1517 1518 rb_link_node(&rs->rs_node, parent, newn); 1519 rb_insert_color(&rs->rs_node, &rgd->rd_rstree); 1520 1521 /* Do our rgrp accounting for the reservation */ 1522 rgd->rd_reserved += rs->rs_free; /* blocks reserved */ 1523 spin_unlock(&rgd->rd_rsspin); 1524 trace_gfs2_rs(rs, TRACE_RS_INSERT); 1525 } 1526 1527 /** 1528 * rgd_free - return the number of free blocks we can allocate. 1529 * @rgd: the resource group 1530 * 1531 * This function returns the number of free blocks for an rgrp. 1532 * That's the clone-free blocks (blocks that are free, not including those 1533 * still being used for unlinked files that haven't been deleted.) 1534 * 1535 * It also subtracts any blocks reserved by someone else, but does not 1536 * include free blocks that are still part of our current reservation, 1537 * because obviously we can (and will) allocate them. 1538 */ 1539 static inline u32 rgd_free(struct gfs2_rgrpd *rgd, struct gfs2_blkreserv *rs) 1540 { 1541 u32 tot_reserved, tot_free; 1542 1543 if (WARN_ON_ONCE(rgd->rd_reserved < rs->rs_free)) 1544 return 0; 1545 tot_reserved = rgd->rd_reserved - rs->rs_free; 1546 1547 if (rgd->rd_free_clone < tot_reserved) 1548 tot_reserved = 0; 1549 1550 tot_free = rgd->rd_free_clone - tot_reserved; 1551 1552 return tot_free; 1553 } 1554 1555 /** 1556 * rg_mblk_search - find a group of multiple free blocks to form a reservation 1557 * @rgd: the resource group descriptor 1558 * @ip: pointer to the inode for which we're reserving blocks 1559 * @ap: the allocation parameters 1560 * 1561 */ 1562 1563 static void rg_mblk_search(struct gfs2_rgrpd *rgd, struct gfs2_inode *ip, 1564 const struct gfs2_alloc_parms *ap) 1565 { 1566 struct gfs2_rbm rbm = { .rgd = rgd, }; 1567 u64 goal; 1568 struct gfs2_blkreserv *rs = &ip->i_res; 1569 u32 extlen; 1570 u32 free_blocks = rgd_free(rgd, rs); 1571 int ret; 1572 struct inode *inode = &ip->i_inode; 1573 1574 if (S_ISDIR(inode->i_mode)) 1575 extlen = 1; 1576 else { 1577 extlen = max_t(u32, atomic_read(&ip->i_sizehint), ap->target); 1578 extlen = clamp(extlen, (u32)RGRP_RSRV_MINBLKS, free_blocks); 1579 } 1580 if ((rgd->rd_free_clone < rgd->rd_reserved) || (free_blocks < extlen)) 1581 return; 1582 1583 /* Find bitmap block that contains bits for goal block */ 1584 if (rgrp_contains_block(rgd, ip->i_goal)) 1585 goal = ip->i_goal; 1586 else 1587 goal = rgd->rd_last_alloc + rgd->rd_data0; 1588 1589 if (WARN_ON(gfs2_rbm_from_block(&rbm, goal))) 1590 return; 1591 1592 ret = gfs2_rbm_find(&rbm, GFS2_BLKST_FREE, &extlen, ip, true); 1593 if (ret == 0) { 1594 rs->rs_rbm = rbm; 1595 rs->rs_free = extlen; 1596 rs_insert(ip); 1597 } else { 1598 if (goal == rgd->rd_last_alloc + rgd->rd_data0) 1599 rgd->rd_last_alloc = 0; 1600 } 1601 } 1602 1603 /** 1604 * gfs2_next_unreserved_block - Return next block that is not reserved 1605 * @rgd: The resource group 1606 * @block: The starting block 1607 * @length: The required length 1608 * @ip: Ignore any reservations for this inode 1609 * 1610 * If the block does not appear in any reservation, then return the 1611 * block number unchanged. If it does appear in the reservation, then 1612 * keep looking through the tree of reservations in order to find the 1613 * first block number which is not reserved. 1614 */ 1615 1616 static u64 gfs2_next_unreserved_block(struct gfs2_rgrpd *rgd, u64 block, 1617 u32 length, 1618 const struct gfs2_inode *ip) 1619 { 1620 struct gfs2_blkreserv *rs; 1621 struct rb_node *n; 1622 int rc; 1623 1624 spin_lock(&rgd->rd_rsspin); 1625 n = rgd->rd_rstree.rb_node; 1626 while (n) { 1627 rs = rb_entry(n, struct gfs2_blkreserv, rs_node); 1628 rc = rs_cmp(block, length, rs); 1629 if (rc < 0) 1630 n = n->rb_left; 1631 else if (rc > 0) 1632 n = n->rb_right; 1633 else 1634 break; 1635 } 1636 1637 if (n) { 1638 while ((rs_cmp(block, length, rs) == 0) && (&ip->i_res != rs)) { 1639 block = gfs2_rbm_to_block(&rs->rs_rbm) + rs->rs_free; 1640 n = n->rb_right; 1641 if (n == NULL) 1642 break; 1643 rs = rb_entry(n, struct gfs2_blkreserv, rs_node); 1644 } 1645 } 1646 1647 spin_unlock(&rgd->rd_rsspin); 1648 return block; 1649 } 1650 1651 /** 1652 * gfs2_reservation_check_and_update - Check for reservations during block alloc 1653 * @rbm: The current position in the resource group 1654 * @ip: The inode for which we are searching for blocks 1655 * @minext: The minimum extent length 1656 * @maxext: A pointer to the maximum extent structure 1657 * 1658 * This checks the current position in the rgrp to see whether there is 1659 * a reservation covering this block. If not then this function is a 1660 * no-op. If there is, then the position is moved to the end of the 1661 * contiguous reservation(s) so that we are pointing at the first 1662 * non-reserved block. 1663 * 1664 * Returns: 0 if no reservation, 1 if @rbm has changed, otherwise an error 1665 */ 1666 1667 static int gfs2_reservation_check_and_update(struct gfs2_rbm *rbm, 1668 const struct gfs2_inode *ip, 1669 u32 minext, 1670 struct gfs2_extent *maxext) 1671 { 1672 u64 block = gfs2_rbm_to_block(rbm); 1673 u32 extlen = 1; 1674 u64 nblock; 1675 int ret; 1676 1677 /* 1678 * If we have a minimum extent length, then skip over any extent 1679 * which is less than the min extent length in size. 1680 */ 1681 if (minext) { 1682 extlen = gfs2_free_extlen(rbm, minext); 1683 if (extlen <= maxext->len) 1684 goto fail; 1685 } 1686 1687 /* 1688 * Check the extent which has been found against the reservations 1689 * and skip if parts of it are already reserved 1690 */ 1691 nblock = gfs2_next_unreserved_block(rbm->rgd, block, extlen, ip); 1692 if (nblock == block) { 1693 if (!minext || extlen >= minext) 1694 return 0; 1695 1696 if (extlen > maxext->len) { 1697 maxext->len = extlen; 1698 maxext->rbm = *rbm; 1699 } 1700 fail: 1701 nblock = block + extlen; 1702 } 1703 ret = gfs2_rbm_from_block(rbm, nblock); 1704 if (ret < 0) 1705 return ret; 1706 return 1; 1707 } 1708 1709 /** 1710 * gfs2_rbm_find - Look for blocks of a particular state 1711 * @rbm: Value/result starting position and final position 1712 * @state: The state which we want to find 1713 * @minext: Pointer to the requested extent length (NULL for a single block) 1714 * This is updated to be the actual reservation size. 1715 * @ip: If set, check for reservations 1716 * @nowrap: Stop looking at the end of the rgrp, rather than wrapping 1717 * around until we've reached the starting point. 1718 * 1719 * Side effects: 1720 * - If looking for free blocks, we set GBF_FULL on each bitmap which 1721 * has no free blocks in it. 1722 * - If looking for free blocks, we set rd_extfail_pt on each rgrp which 1723 * has come up short on a free block search. 1724 * 1725 * Returns: 0 on success, -ENOSPC if there is no block of the requested state 1726 */ 1727 1728 static int gfs2_rbm_find(struct gfs2_rbm *rbm, u8 state, u32 *minext, 1729 const struct gfs2_inode *ip, bool nowrap) 1730 { 1731 struct buffer_head *bh; 1732 int initial_bii; 1733 u32 initial_offset; 1734 int first_bii = rbm->bii; 1735 u32 first_offset = rbm->offset; 1736 u32 offset; 1737 u8 *buffer; 1738 int n = 0; 1739 int iters = rbm->rgd->rd_length; 1740 int ret; 1741 struct gfs2_bitmap *bi; 1742 struct gfs2_extent maxext = { .rbm.rgd = rbm->rgd, }; 1743 1744 /* If we are not starting at the beginning of a bitmap, then we 1745 * need to add one to the bitmap count to ensure that we search 1746 * the starting bitmap twice. 1747 */ 1748 if (rbm->offset != 0) 1749 iters++; 1750 1751 while(1) { 1752 bi = rbm_bi(rbm); 1753 if ((ip == NULL || !gfs2_rs_active(&ip->i_res)) && 1754 test_bit(GBF_FULL, &bi->bi_flags) && 1755 (state == GFS2_BLKST_FREE)) 1756 goto next_bitmap; 1757 1758 bh = bi->bi_bh; 1759 buffer = bh->b_data + bi->bi_offset; 1760 WARN_ON(!buffer_uptodate(bh)); 1761 if (state != GFS2_BLKST_UNLINKED && bi->bi_clone) 1762 buffer = bi->bi_clone + bi->bi_offset; 1763 initial_offset = rbm->offset; 1764 offset = gfs2_bitfit(buffer, bi->bi_bytes, rbm->offset, state); 1765 if (offset == BFITNOENT) 1766 goto bitmap_full; 1767 rbm->offset = offset; 1768 if (ip == NULL) 1769 return 0; 1770 1771 initial_bii = rbm->bii; 1772 ret = gfs2_reservation_check_and_update(rbm, ip, 1773 minext ? *minext : 0, 1774 &maxext); 1775 if (ret == 0) 1776 return 0; 1777 if (ret > 0) { 1778 n += (rbm->bii - initial_bii); 1779 goto next_iter; 1780 } 1781 if (ret == -E2BIG) { 1782 rbm->bii = 0; 1783 rbm->offset = 0; 1784 n += (rbm->bii - initial_bii); 1785 goto res_covered_end_of_rgrp; 1786 } 1787 return ret; 1788 1789 bitmap_full: /* Mark bitmap as full and fall through */ 1790 if ((state == GFS2_BLKST_FREE) && initial_offset == 0) 1791 set_bit(GBF_FULL, &bi->bi_flags); 1792 1793 next_bitmap: /* Find next bitmap in the rgrp */ 1794 rbm->offset = 0; 1795 rbm->bii++; 1796 if (rbm->bii == rbm->rgd->rd_length) 1797 rbm->bii = 0; 1798 res_covered_end_of_rgrp: 1799 if ((rbm->bii == 0) && nowrap) 1800 break; 1801 n++; 1802 next_iter: 1803 if (n >= iters) 1804 break; 1805 } 1806 1807 if (minext == NULL || state != GFS2_BLKST_FREE) 1808 return -ENOSPC; 1809 1810 /* If the extent was too small, and it's smaller than the smallest 1811 to have failed before, remember for future reference that it's 1812 useless to search this rgrp again for this amount or more. */ 1813 if ((first_offset == 0) && (first_bii == 0) && 1814 (*minext < rbm->rgd->rd_extfail_pt)) 1815 rbm->rgd->rd_extfail_pt = *minext; 1816 1817 /* If the maximum extent we found is big enough to fulfill the 1818 minimum requirements, use it anyway. */ 1819 if (maxext.len) { 1820 *rbm = maxext.rbm; 1821 *minext = maxext.len; 1822 return 0; 1823 } 1824 1825 return -ENOSPC; 1826 } 1827 1828 /** 1829 * try_rgrp_unlink - Look for any unlinked, allocated, but unused inodes 1830 * @rgd: The rgrp 1831 * @last_unlinked: block address of the last dinode we unlinked 1832 * @skip: block address we should explicitly not unlink 1833 * 1834 * Returns: 0 if no error 1835 * The inode, if one has been found, in inode. 1836 */ 1837 1838 static void try_rgrp_unlink(struct gfs2_rgrpd *rgd, u64 *last_unlinked, u64 skip) 1839 { 1840 u64 block; 1841 struct gfs2_sbd *sdp = rgd->rd_sbd; 1842 struct gfs2_glock *gl; 1843 struct gfs2_inode *ip; 1844 int error; 1845 int found = 0; 1846 struct gfs2_rbm rbm = { .rgd = rgd, .bii = 0, .offset = 0 }; 1847 1848 while (1) { 1849 down_write(&sdp->sd_log_flush_lock); 1850 error = gfs2_rbm_find(&rbm, GFS2_BLKST_UNLINKED, NULL, NULL, 1851 true); 1852 up_write(&sdp->sd_log_flush_lock); 1853 if (error == -ENOSPC) 1854 break; 1855 if (WARN_ON_ONCE(error)) 1856 break; 1857 1858 block = gfs2_rbm_to_block(&rbm); 1859 if (gfs2_rbm_from_block(&rbm, block + 1)) 1860 break; 1861 if (*last_unlinked != NO_BLOCK && block <= *last_unlinked) 1862 continue; 1863 if (block == skip) 1864 continue; 1865 *last_unlinked = block; 1866 1867 error = gfs2_glock_get(sdp, block, &gfs2_iopen_glops, CREATE, &gl); 1868 if (error) 1869 continue; 1870 1871 /* If the inode is already in cache, we can ignore it here 1872 * because the existing inode disposal code will deal with 1873 * it when all refs have gone away. Accessing gl_object like 1874 * this is not safe in general. Here it is ok because we do 1875 * not dereference the pointer, and we only need an approx 1876 * answer to whether it is NULL or not. 1877 */ 1878 ip = gl->gl_object; 1879 1880 if (ip || queue_work(gfs2_delete_workqueue, &gl->gl_delete) == 0) 1881 gfs2_glock_put(gl); 1882 else 1883 found++; 1884 1885 /* Limit reclaim to sensible number of tasks */ 1886 if (found > NR_CPUS) 1887 return; 1888 } 1889 1890 rgd->rd_flags &= ~GFS2_RDF_CHECK; 1891 return; 1892 } 1893 1894 /** 1895 * gfs2_rgrp_congested - Use stats to figure out whether an rgrp is congested 1896 * @rgd: The rgrp in question 1897 * @loops: An indication of how picky we can be (0=very, 1=less so) 1898 * 1899 * This function uses the recently added glock statistics in order to 1900 * figure out whether a parciular resource group is suffering from 1901 * contention from multiple nodes. This is done purely on the basis 1902 * of timings, since this is the only data we have to work with and 1903 * our aim here is to reject a resource group which is highly contended 1904 * but (very important) not to do this too often in order to ensure that 1905 * we do not land up introducing fragmentation by changing resource 1906 * groups when not actually required. 1907 * 1908 * The calculation is fairly simple, we want to know whether the SRTTB 1909 * (i.e. smoothed round trip time for blocking operations) to acquire 1910 * the lock for this rgrp's glock is significantly greater than the 1911 * time taken for resource groups on average. We introduce a margin in 1912 * the form of the variable @var which is computed as the sum of the two 1913 * respective variences, and multiplied by a factor depending on @loops 1914 * and whether we have a lot of data to base the decision on. This is 1915 * then tested against the square difference of the means in order to 1916 * decide whether the result is statistically significant or not. 1917 * 1918 * Returns: A boolean verdict on the congestion status 1919 */ 1920 1921 static bool gfs2_rgrp_congested(const struct gfs2_rgrpd *rgd, int loops) 1922 { 1923 const struct gfs2_glock *gl = rgd->rd_gl; 1924 const struct gfs2_sbd *sdp = gl->gl_name.ln_sbd; 1925 struct gfs2_lkstats *st; 1926 u64 r_dcount, l_dcount; 1927 u64 l_srttb, a_srttb = 0; 1928 s64 srttb_diff; 1929 u64 sqr_diff; 1930 u64 var; 1931 int cpu, nonzero = 0; 1932 1933 preempt_disable(); 1934 for_each_present_cpu(cpu) { 1935 st = &per_cpu_ptr(sdp->sd_lkstats, cpu)->lkstats[LM_TYPE_RGRP]; 1936 if (st->stats[GFS2_LKS_SRTTB]) { 1937 a_srttb += st->stats[GFS2_LKS_SRTTB]; 1938 nonzero++; 1939 } 1940 } 1941 st = &this_cpu_ptr(sdp->sd_lkstats)->lkstats[LM_TYPE_RGRP]; 1942 if (nonzero) 1943 do_div(a_srttb, nonzero); 1944 r_dcount = st->stats[GFS2_LKS_DCOUNT]; 1945 var = st->stats[GFS2_LKS_SRTTVARB] + 1946 gl->gl_stats.stats[GFS2_LKS_SRTTVARB]; 1947 preempt_enable(); 1948 1949 l_srttb = gl->gl_stats.stats[GFS2_LKS_SRTTB]; 1950 l_dcount = gl->gl_stats.stats[GFS2_LKS_DCOUNT]; 1951 1952 if ((l_dcount < 1) || (r_dcount < 1) || (a_srttb == 0)) 1953 return false; 1954 1955 srttb_diff = a_srttb - l_srttb; 1956 sqr_diff = srttb_diff * srttb_diff; 1957 1958 var *= 2; 1959 if (l_dcount < 8 || r_dcount < 8) 1960 var *= 2; 1961 if (loops == 1) 1962 var *= 2; 1963 1964 return ((srttb_diff < 0) && (sqr_diff > var)); 1965 } 1966 1967 /** 1968 * gfs2_rgrp_used_recently 1969 * @rs: The block reservation with the rgrp to test 1970 * @msecs: The time limit in milliseconds 1971 * 1972 * Returns: True if the rgrp glock has been used within the time limit 1973 */ 1974 static bool gfs2_rgrp_used_recently(const struct gfs2_blkreserv *rs, 1975 u64 msecs) 1976 { 1977 u64 tdiff; 1978 1979 tdiff = ktime_to_ns(ktime_sub(ktime_get_real(), 1980 rs->rs_rbm.rgd->rd_gl->gl_dstamp)); 1981 1982 return tdiff > (msecs * 1000 * 1000); 1983 } 1984 1985 static u32 gfs2_orlov_skip(const struct gfs2_inode *ip) 1986 { 1987 const struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 1988 u32 skip; 1989 1990 get_random_bytes(&skip, sizeof(skip)); 1991 return skip % sdp->sd_rgrps; 1992 } 1993 1994 static bool gfs2_select_rgrp(struct gfs2_rgrpd **pos, const struct gfs2_rgrpd *begin) 1995 { 1996 struct gfs2_rgrpd *rgd = *pos; 1997 struct gfs2_sbd *sdp = rgd->rd_sbd; 1998 1999 rgd = gfs2_rgrpd_get_next(rgd); 2000 if (rgd == NULL) 2001 rgd = gfs2_rgrpd_get_first(sdp); 2002 *pos = rgd; 2003 if (rgd != begin) /* If we didn't wrap */ 2004 return true; 2005 return false; 2006 } 2007 2008 /** 2009 * fast_to_acquire - determine if a resource group will be fast to acquire 2010 * 2011 * If this is one of our preferred rgrps, it should be quicker to acquire, 2012 * because we tried to set ourselves up as dlm lock master. 2013 */ 2014 static inline int fast_to_acquire(struct gfs2_rgrpd *rgd) 2015 { 2016 struct gfs2_glock *gl = rgd->rd_gl; 2017 2018 if (gl->gl_state != LM_ST_UNLOCKED && list_empty(&gl->gl_holders) && 2019 !test_bit(GLF_DEMOTE_IN_PROGRESS, &gl->gl_flags) && 2020 !test_bit(GLF_DEMOTE, &gl->gl_flags)) 2021 return 1; 2022 if (rgd->rd_flags & GFS2_RDF_PREFERRED) 2023 return 1; 2024 return 0; 2025 } 2026 2027 /** 2028 * gfs2_inplace_reserve - Reserve space in the filesystem 2029 * @ip: the inode to reserve space for 2030 * @ap: the allocation parameters 2031 * 2032 * We try our best to find an rgrp that has at least ap->target blocks 2033 * available. After a couple of passes (loops == 2), the prospects of finding 2034 * such an rgrp diminish. At this stage, we return the first rgrp that has 2035 * at least ap->min_target blocks available. Either way, we set ap->allowed to 2036 * the number of blocks available in the chosen rgrp. 2037 * 2038 * Returns: 0 on success, 2039 * -ENOMEM if a suitable rgrp can't be found 2040 * errno otherwise 2041 */ 2042 2043 int gfs2_inplace_reserve(struct gfs2_inode *ip, struct gfs2_alloc_parms *ap) 2044 { 2045 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 2046 struct gfs2_rgrpd *begin = NULL; 2047 struct gfs2_blkreserv *rs = &ip->i_res; 2048 int error = 0, rg_locked, flags = 0; 2049 u64 last_unlinked = NO_BLOCK; 2050 int loops = 0; 2051 u32 free_blocks, skip = 0; 2052 2053 if (sdp->sd_args.ar_rgrplvb) 2054 flags |= GL_SKIP; 2055 if (gfs2_assert_warn(sdp, ap->target)) 2056 return -EINVAL; 2057 if (gfs2_rs_active(rs)) { 2058 begin = rs->rs_rbm.rgd; 2059 } else if (rs->rs_rbm.rgd && 2060 rgrp_contains_block(rs->rs_rbm.rgd, ip->i_goal)) { 2061 begin = rs->rs_rbm.rgd; 2062 } else { 2063 check_and_update_goal(ip); 2064 rs->rs_rbm.rgd = begin = gfs2_blk2rgrpd(sdp, ip->i_goal, 1); 2065 } 2066 if (S_ISDIR(ip->i_inode.i_mode) && (ap->aflags & GFS2_AF_ORLOV)) 2067 skip = gfs2_orlov_skip(ip); 2068 if (rs->rs_rbm.rgd == NULL) 2069 return -EBADSLT; 2070 2071 while (loops < 3) { 2072 rg_locked = 1; 2073 2074 if (!gfs2_glock_is_locked_by_me(rs->rs_rbm.rgd->rd_gl)) { 2075 rg_locked = 0; 2076 if (skip && skip--) 2077 goto next_rgrp; 2078 if (!gfs2_rs_active(rs)) { 2079 if (loops == 0 && 2080 !fast_to_acquire(rs->rs_rbm.rgd)) 2081 goto next_rgrp; 2082 if ((loops < 2) && 2083 gfs2_rgrp_used_recently(rs, 1000) && 2084 gfs2_rgrp_congested(rs->rs_rbm.rgd, loops)) 2085 goto next_rgrp; 2086 } 2087 error = gfs2_glock_nq_init(rs->rs_rbm.rgd->rd_gl, 2088 LM_ST_EXCLUSIVE, flags, 2089 &ip->i_rgd_gh); 2090 if (unlikely(error)) 2091 return error; 2092 if (!gfs2_rs_active(rs) && (loops < 2) && 2093 gfs2_rgrp_congested(rs->rs_rbm.rgd, loops)) 2094 goto skip_rgrp; 2095 if (sdp->sd_args.ar_rgrplvb) { 2096 error = update_rgrp_lvb(rs->rs_rbm.rgd); 2097 if (unlikely(error)) { 2098 gfs2_glock_dq_uninit(&ip->i_rgd_gh); 2099 return error; 2100 } 2101 } 2102 } 2103 2104 /* Skip unusable resource groups */ 2105 if ((rs->rs_rbm.rgd->rd_flags & (GFS2_RGF_NOALLOC | 2106 GFS2_RDF_ERROR)) || 2107 (loops == 0 && ap->target > rs->rs_rbm.rgd->rd_extfail_pt)) 2108 goto skip_rgrp; 2109 2110 if (sdp->sd_args.ar_rgrplvb) 2111 gfs2_rgrp_bh_get(rs->rs_rbm.rgd); 2112 2113 /* Get a reservation if we don't already have one */ 2114 if (!gfs2_rs_active(rs)) 2115 rg_mblk_search(rs->rs_rbm.rgd, ip, ap); 2116 2117 /* Skip rgrps when we can't get a reservation on first pass */ 2118 if (!gfs2_rs_active(rs) && (loops < 1)) 2119 goto check_rgrp; 2120 2121 /* If rgrp has enough free space, use it */ 2122 free_blocks = rgd_free(rs->rs_rbm.rgd, rs); 2123 if (free_blocks >= ap->target || 2124 (loops == 2 && ap->min_target && 2125 free_blocks >= ap->min_target)) { 2126 ap->allowed = free_blocks; 2127 return 0; 2128 } 2129 check_rgrp: 2130 /* Check for unlinked inodes which can be reclaimed */ 2131 if (rs->rs_rbm.rgd->rd_flags & GFS2_RDF_CHECK) 2132 try_rgrp_unlink(rs->rs_rbm.rgd, &last_unlinked, 2133 ip->i_no_addr); 2134 skip_rgrp: 2135 /* Drop reservation, if we couldn't use reserved rgrp */ 2136 if (gfs2_rs_active(rs)) 2137 gfs2_rs_deltree(rs); 2138 2139 /* Unlock rgrp if required */ 2140 if (!rg_locked) 2141 gfs2_glock_dq_uninit(&ip->i_rgd_gh); 2142 next_rgrp: 2143 /* Find the next rgrp, and continue looking */ 2144 if (gfs2_select_rgrp(&rs->rs_rbm.rgd, begin)) 2145 continue; 2146 if (skip) 2147 continue; 2148 2149 /* If we've scanned all the rgrps, but found no free blocks 2150 * then this checks for some less likely conditions before 2151 * trying again. 2152 */ 2153 loops++; 2154 /* Check that fs hasn't grown if writing to rindex */ 2155 if (ip == GFS2_I(sdp->sd_rindex) && !sdp->sd_rindex_uptodate) { 2156 error = gfs2_ri_update(ip); 2157 if (error) 2158 return error; 2159 } 2160 /* Flushing the log may release space */ 2161 if (loops == 2) 2162 gfs2_log_flush(sdp, NULL, GFS2_LOG_HEAD_FLUSH_NORMAL | 2163 GFS2_LFC_INPLACE_RESERVE); 2164 } 2165 2166 return -ENOSPC; 2167 } 2168 2169 /** 2170 * gfs2_inplace_release - release an inplace reservation 2171 * @ip: the inode the reservation was taken out on 2172 * 2173 * Release a reservation made by gfs2_inplace_reserve(). 2174 */ 2175 2176 void gfs2_inplace_release(struct gfs2_inode *ip) 2177 { 2178 if (gfs2_holder_initialized(&ip->i_rgd_gh)) 2179 gfs2_glock_dq_uninit(&ip->i_rgd_gh); 2180 } 2181 2182 /** 2183 * gfs2_alloc_extent - allocate an extent from a given bitmap 2184 * @rbm: the resource group information 2185 * @dinode: TRUE if the first block we allocate is for a dinode 2186 * @n: The extent length (value/result) 2187 * 2188 * Add the bitmap buffer to the transaction. 2189 * Set the found bits to @new_state to change block's allocation state. 2190 */ 2191 static void gfs2_alloc_extent(const struct gfs2_rbm *rbm, bool dinode, 2192 unsigned int *n) 2193 { 2194 struct gfs2_rbm pos = { .rgd = rbm->rgd, }; 2195 const unsigned int elen = *n; 2196 u64 block; 2197 int ret; 2198 2199 *n = 1; 2200 block = gfs2_rbm_to_block(rbm); 2201 gfs2_trans_add_meta(rbm->rgd->rd_gl, rbm_bi(rbm)->bi_bh); 2202 gfs2_setbit(rbm, true, dinode ? GFS2_BLKST_DINODE : GFS2_BLKST_USED); 2203 block++; 2204 while (*n < elen) { 2205 ret = gfs2_rbm_from_block(&pos, block); 2206 if (ret || gfs2_testbit(&pos, true) != GFS2_BLKST_FREE) 2207 break; 2208 gfs2_trans_add_meta(pos.rgd->rd_gl, rbm_bi(&pos)->bi_bh); 2209 gfs2_setbit(&pos, true, GFS2_BLKST_USED); 2210 (*n)++; 2211 block++; 2212 } 2213 } 2214 2215 /** 2216 * rgblk_free - Change alloc state of given block(s) 2217 * @sdp: the filesystem 2218 * @rgd: the resource group the blocks are in 2219 * @bstart: the start of a run of blocks to free 2220 * @blen: the length of the block run (all must lie within ONE RG!) 2221 * @new_state: GFS2_BLKST_XXX the after-allocation block state 2222 */ 2223 2224 static void rgblk_free(struct gfs2_sbd *sdp, struct gfs2_rgrpd *rgd, 2225 u64 bstart, u32 blen, unsigned char new_state) 2226 { 2227 struct gfs2_rbm rbm; 2228 struct gfs2_bitmap *bi, *bi_prev = NULL; 2229 2230 rbm.rgd = rgd; 2231 if (WARN_ON_ONCE(gfs2_rbm_from_block(&rbm, bstart))) 2232 return; 2233 while (blen--) { 2234 bi = rbm_bi(&rbm); 2235 if (bi != bi_prev) { 2236 if (!bi->bi_clone) { 2237 bi->bi_clone = kmalloc(bi->bi_bh->b_size, 2238 GFP_NOFS | __GFP_NOFAIL); 2239 memcpy(bi->bi_clone + bi->bi_offset, 2240 bi->bi_bh->b_data + bi->bi_offset, 2241 bi->bi_bytes); 2242 } 2243 gfs2_trans_add_meta(rbm.rgd->rd_gl, bi->bi_bh); 2244 bi_prev = bi; 2245 } 2246 gfs2_setbit(&rbm, false, new_state); 2247 gfs2_rbm_incr(&rbm); 2248 } 2249 } 2250 2251 /** 2252 * gfs2_rgrp_dump - print out an rgrp 2253 * @seq: The iterator 2254 * @gl: The glock in question 2255 * 2256 */ 2257 2258 void gfs2_rgrp_dump(struct seq_file *seq, const struct gfs2_glock *gl) 2259 { 2260 struct gfs2_rgrpd *rgd = gl->gl_object; 2261 struct gfs2_blkreserv *trs; 2262 const struct rb_node *n; 2263 2264 if (rgd == NULL) 2265 return; 2266 gfs2_print_dbg(seq, " R: n:%llu f:%02x b:%u/%u i:%u r:%u e:%u\n", 2267 (unsigned long long)rgd->rd_addr, rgd->rd_flags, 2268 rgd->rd_free, rgd->rd_free_clone, rgd->rd_dinodes, 2269 rgd->rd_reserved, rgd->rd_extfail_pt); 2270 if (rgd->rd_sbd->sd_args.ar_rgrplvb) { 2271 struct gfs2_rgrp_lvb *rgl = rgd->rd_rgl; 2272 2273 gfs2_print_dbg(seq, " L: f:%02x b:%u i:%u\n", 2274 be32_to_cpu(rgl->rl_flags), 2275 be32_to_cpu(rgl->rl_free), 2276 be32_to_cpu(rgl->rl_dinodes)); 2277 } 2278 spin_lock(&rgd->rd_rsspin); 2279 for (n = rb_first(&rgd->rd_rstree); n; n = rb_next(&trs->rs_node)) { 2280 trs = rb_entry(n, struct gfs2_blkreserv, rs_node); 2281 dump_rs(seq, trs); 2282 } 2283 spin_unlock(&rgd->rd_rsspin); 2284 } 2285 2286 static void gfs2_rgrp_error(struct gfs2_rgrpd *rgd) 2287 { 2288 struct gfs2_sbd *sdp = rgd->rd_sbd; 2289 fs_warn(sdp, "rgrp %llu has an error, marking it readonly until umount\n", 2290 (unsigned long long)rgd->rd_addr); 2291 fs_warn(sdp, "umount on all nodes and run fsck.gfs2 to fix the error\n"); 2292 gfs2_rgrp_dump(NULL, rgd->rd_gl); 2293 rgd->rd_flags |= GFS2_RDF_ERROR; 2294 } 2295 2296 /** 2297 * gfs2_adjust_reservation - Adjust (or remove) a reservation after allocation 2298 * @ip: The inode we have just allocated blocks for 2299 * @rbm: The start of the allocated blocks 2300 * @len: The extent length 2301 * 2302 * Adjusts a reservation after an allocation has taken place. If the 2303 * reservation does not match the allocation, or if it is now empty 2304 * then it is removed. 2305 */ 2306 2307 static void gfs2_adjust_reservation(struct gfs2_inode *ip, 2308 const struct gfs2_rbm *rbm, unsigned len) 2309 { 2310 struct gfs2_blkreserv *rs = &ip->i_res; 2311 struct gfs2_rgrpd *rgd = rbm->rgd; 2312 unsigned rlen; 2313 u64 block; 2314 int ret; 2315 2316 spin_lock(&rgd->rd_rsspin); 2317 if (gfs2_rs_active(rs)) { 2318 if (gfs2_rbm_eq(&rs->rs_rbm, rbm)) { 2319 block = gfs2_rbm_to_block(rbm); 2320 ret = gfs2_rbm_from_block(&rs->rs_rbm, block + len); 2321 rlen = min(rs->rs_free, len); 2322 rs->rs_free -= rlen; 2323 rgd->rd_reserved -= rlen; 2324 trace_gfs2_rs(rs, TRACE_RS_CLAIM); 2325 if (rs->rs_free && !ret) 2326 goto out; 2327 /* We used up our block reservation, so we should 2328 reserve more blocks next time. */ 2329 atomic_add(RGRP_RSRV_ADDBLKS, &ip->i_sizehint); 2330 } 2331 __rs_deltree(rs); 2332 } 2333 out: 2334 spin_unlock(&rgd->rd_rsspin); 2335 } 2336 2337 /** 2338 * gfs2_set_alloc_start - Set starting point for block allocation 2339 * @rbm: The rbm which will be set to the required location 2340 * @ip: The gfs2 inode 2341 * @dinode: Flag to say if allocation includes a new inode 2342 * 2343 * This sets the starting point from the reservation if one is active 2344 * otherwise it falls back to guessing a start point based on the 2345 * inode's goal block or the last allocation point in the rgrp. 2346 */ 2347 2348 static void gfs2_set_alloc_start(struct gfs2_rbm *rbm, 2349 const struct gfs2_inode *ip, bool dinode) 2350 { 2351 u64 goal; 2352 2353 if (gfs2_rs_active(&ip->i_res)) { 2354 *rbm = ip->i_res.rs_rbm; 2355 return; 2356 } 2357 2358 if (!dinode && rgrp_contains_block(rbm->rgd, ip->i_goal)) 2359 goal = ip->i_goal; 2360 else 2361 goal = rbm->rgd->rd_last_alloc + rbm->rgd->rd_data0; 2362 2363 if (WARN_ON_ONCE(gfs2_rbm_from_block(rbm, goal))) { 2364 rbm->bii = 0; 2365 rbm->offset = 0; 2366 } 2367 } 2368 2369 /** 2370 * gfs2_alloc_blocks - Allocate one or more blocks of data and/or a dinode 2371 * @ip: the inode to allocate the block for 2372 * @bn: Used to return the starting block number 2373 * @nblocks: requested number of blocks/extent length (value/result) 2374 * @dinode: 1 if we're allocating a dinode block, else 0 2375 * @generation: the generation number of the inode 2376 * 2377 * Returns: 0 or error 2378 */ 2379 2380 int gfs2_alloc_blocks(struct gfs2_inode *ip, u64 *bn, unsigned int *nblocks, 2381 bool dinode, u64 *generation) 2382 { 2383 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 2384 struct buffer_head *dibh; 2385 struct gfs2_rbm rbm = { .rgd = ip->i_res.rs_rbm.rgd, }; 2386 unsigned int ndata; 2387 u64 block; /* block, within the file system scope */ 2388 int error; 2389 2390 gfs2_set_alloc_start(&rbm, ip, dinode); 2391 error = gfs2_rbm_find(&rbm, GFS2_BLKST_FREE, NULL, ip, false); 2392 2393 if (error == -ENOSPC) { 2394 gfs2_set_alloc_start(&rbm, ip, dinode); 2395 error = gfs2_rbm_find(&rbm, GFS2_BLKST_FREE, NULL, NULL, false); 2396 } 2397 2398 /* Since all blocks are reserved in advance, this shouldn't happen */ 2399 if (error) { 2400 fs_warn(sdp, "inum=%llu error=%d, nblocks=%u, full=%d fail_pt=%d\n", 2401 (unsigned long long)ip->i_no_addr, error, *nblocks, 2402 test_bit(GBF_FULL, &rbm.rgd->rd_bits->bi_flags), 2403 rbm.rgd->rd_extfail_pt); 2404 goto rgrp_error; 2405 } 2406 2407 gfs2_alloc_extent(&rbm, dinode, nblocks); 2408 block = gfs2_rbm_to_block(&rbm); 2409 rbm.rgd->rd_last_alloc = block - rbm.rgd->rd_data0; 2410 if (gfs2_rs_active(&ip->i_res)) 2411 gfs2_adjust_reservation(ip, &rbm, *nblocks); 2412 ndata = *nblocks; 2413 if (dinode) 2414 ndata--; 2415 2416 if (!dinode) { 2417 ip->i_goal = block + ndata - 1; 2418 error = gfs2_meta_inode_buffer(ip, &dibh); 2419 if (error == 0) { 2420 struct gfs2_dinode *di = 2421 (struct gfs2_dinode *)dibh->b_data; 2422 gfs2_trans_add_meta(ip->i_gl, dibh); 2423 di->di_goal_meta = di->di_goal_data = 2424 cpu_to_be64(ip->i_goal); 2425 brelse(dibh); 2426 } 2427 } 2428 if (rbm.rgd->rd_free < *nblocks) { 2429 fs_warn(sdp, "nblocks=%u\n", *nblocks); 2430 goto rgrp_error; 2431 } 2432 2433 rbm.rgd->rd_free -= *nblocks; 2434 if (dinode) { 2435 rbm.rgd->rd_dinodes++; 2436 *generation = rbm.rgd->rd_igeneration++; 2437 if (*generation == 0) 2438 *generation = rbm.rgd->rd_igeneration++; 2439 } 2440 2441 gfs2_trans_add_meta(rbm.rgd->rd_gl, rbm.rgd->rd_bits[0].bi_bh); 2442 gfs2_rgrp_out(rbm.rgd, rbm.rgd->rd_bits[0].bi_bh->b_data); 2443 2444 gfs2_statfs_change(sdp, 0, -(s64)*nblocks, dinode ? 1 : 0); 2445 if (dinode) 2446 gfs2_trans_add_unrevoke(sdp, block, *nblocks); 2447 2448 gfs2_quota_change(ip, *nblocks, ip->i_inode.i_uid, ip->i_inode.i_gid); 2449 2450 rbm.rgd->rd_free_clone -= *nblocks; 2451 trace_gfs2_block_alloc(ip, rbm.rgd, block, *nblocks, 2452 dinode ? GFS2_BLKST_DINODE : GFS2_BLKST_USED); 2453 *bn = block; 2454 return 0; 2455 2456 rgrp_error: 2457 gfs2_rgrp_error(rbm.rgd); 2458 return -EIO; 2459 } 2460 2461 /** 2462 * __gfs2_free_blocks - free a contiguous run of block(s) 2463 * @ip: the inode these blocks are being freed from 2464 * @rgd: the resource group the blocks are in 2465 * @bstart: first block of a run of contiguous blocks 2466 * @blen: the length of the block run 2467 * @meta: 1 if the blocks represent metadata 2468 * 2469 */ 2470 2471 void __gfs2_free_blocks(struct gfs2_inode *ip, struct gfs2_rgrpd *rgd, 2472 u64 bstart, u32 blen, int meta) 2473 { 2474 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 2475 2476 rgblk_free(sdp, rgd, bstart, blen, GFS2_BLKST_FREE); 2477 trace_gfs2_block_alloc(ip, rgd, bstart, blen, GFS2_BLKST_FREE); 2478 rgd->rd_free += blen; 2479 rgd->rd_flags &= ~GFS2_RGF_TRIMMED; 2480 gfs2_trans_add_meta(rgd->rd_gl, rgd->rd_bits[0].bi_bh); 2481 gfs2_rgrp_out(rgd, rgd->rd_bits[0].bi_bh->b_data); 2482 2483 /* Directories keep their data in the metadata address space */ 2484 if (meta || ip->i_depth) 2485 gfs2_meta_wipe(ip, bstart, blen); 2486 } 2487 2488 /** 2489 * gfs2_free_meta - free a contiguous run of data block(s) 2490 * @ip: the inode these blocks are being freed from 2491 * @rgd: the resource group the blocks are in 2492 * @bstart: first block of a run of contiguous blocks 2493 * @blen: the length of the block run 2494 * 2495 */ 2496 2497 void gfs2_free_meta(struct gfs2_inode *ip, struct gfs2_rgrpd *rgd, 2498 u64 bstart, u32 blen) 2499 { 2500 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 2501 2502 __gfs2_free_blocks(ip, rgd, bstart, blen, 1); 2503 gfs2_statfs_change(sdp, 0, +blen, 0); 2504 gfs2_quota_change(ip, -(s64)blen, ip->i_inode.i_uid, ip->i_inode.i_gid); 2505 } 2506 2507 void gfs2_unlink_di(struct inode *inode) 2508 { 2509 struct gfs2_inode *ip = GFS2_I(inode); 2510 struct gfs2_sbd *sdp = GFS2_SB(inode); 2511 struct gfs2_rgrpd *rgd; 2512 u64 blkno = ip->i_no_addr; 2513 2514 rgd = gfs2_blk2rgrpd(sdp, blkno, true); 2515 if (!rgd) 2516 return; 2517 rgblk_free(sdp, rgd, blkno, 1, GFS2_BLKST_UNLINKED); 2518 trace_gfs2_block_alloc(ip, rgd, blkno, 1, GFS2_BLKST_UNLINKED); 2519 gfs2_trans_add_meta(rgd->rd_gl, rgd->rd_bits[0].bi_bh); 2520 gfs2_rgrp_out(rgd, rgd->rd_bits[0].bi_bh->b_data); 2521 be32_add_cpu(&rgd->rd_rgl->rl_unlinked, 1); 2522 } 2523 2524 void gfs2_free_di(struct gfs2_rgrpd *rgd, struct gfs2_inode *ip) 2525 { 2526 struct gfs2_sbd *sdp = rgd->rd_sbd; 2527 2528 rgblk_free(sdp, rgd, ip->i_no_addr, 1, GFS2_BLKST_FREE); 2529 if (!rgd->rd_dinodes) 2530 gfs2_consist_rgrpd(rgd); 2531 rgd->rd_dinodes--; 2532 rgd->rd_free++; 2533 2534 gfs2_trans_add_meta(rgd->rd_gl, rgd->rd_bits[0].bi_bh); 2535 gfs2_rgrp_out(rgd, rgd->rd_bits[0].bi_bh->b_data); 2536 be32_add_cpu(&rgd->rd_rgl->rl_unlinked, -1); 2537 2538 gfs2_statfs_change(sdp, 0, +1, -1); 2539 trace_gfs2_block_alloc(ip, rgd, ip->i_no_addr, 1, GFS2_BLKST_FREE); 2540 gfs2_quota_change(ip, -1, ip->i_inode.i_uid, ip->i_inode.i_gid); 2541 gfs2_meta_wipe(ip, ip->i_no_addr, 1); 2542 } 2543 2544 /** 2545 * gfs2_check_blk_type - Check the type of a block 2546 * @sdp: The superblock 2547 * @no_addr: The block number to check 2548 * @type: The block type we are looking for 2549 * 2550 * Returns: 0 if the block type matches the expected type 2551 * -ESTALE if it doesn't match 2552 * or -ve errno if something went wrong while checking 2553 */ 2554 2555 int gfs2_check_blk_type(struct gfs2_sbd *sdp, u64 no_addr, unsigned int type) 2556 { 2557 struct gfs2_rgrpd *rgd; 2558 struct gfs2_holder rgd_gh; 2559 struct gfs2_rbm rbm; 2560 int error = -EINVAL; 2561 2562 rgd = gfs2_blk2rgrpd(sdp, no_addr, 1); 2563 if (!rgd) 2564 goto fail; 2565 2566 error = gfs2_glock_nq_init(rgd->rd_gl, LM_ST_SHARED, 0, &rgd_gh); 2567 if (error) 2568 goto fail; 2569 2570 rbm.rgd = rgd; 2571 error = gfs2_rbm_from_block(&rbm, no_addr); 2572 if (WARN_ON_ONCE(error)) 2573 goto fail; 2574 2575 if (gfs2_testbit(&rbm, false) != type) 2576 error = -ESTALE; 2577 2578 gfs2_glock_dq_uninit(&rgd_gh); 2579 fail: 2580 return error; 2581 } 2582 2583 /** 2584 * gfs2_rlist_add - add a RG to a list of RGs 2585 * @ip: the inode 2586 * @rlist: the list of resource groups 2587 * @block: the block 2588 * 2589 * Figure out what RG a block belongs to and add that RG to the list 2590 * 2591 * FIXME: Don't use NOFAIL 2592 * 2593 */ 2594 2595 void gfs2_rlist_add(struct gfs2_inode *ip, struct gfs2_rgrp_list *rlist, 2596 u64 block) 2597 { 2598 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 2599 struct gfs2_rgrpd *rgd; 2600 struct gfs2_rgrpd **tmp; 2601 unsigned int new_space; 2602 unsigned int x; 2603 2604 if (gfs2_assert_warn(sdp, !rlist->rl_ghs)) 2605 return; 2606 2607 /* 2608 * The resource group last accessed is kept in the last position. 2609 */ 2610 2611 if (rlist->rl_rgrps) { 2612 rgd = rlist->rl_rgd[rlist->rl_rgrps - 1]; 2613 if (rgrp_contains_block(rgd, block)) 2614 return; 2615 rgd = gfs2_blk2rgrpd(sdp, block, 1); 2616 } else { 2617 rgd = ip->i_res.rs_rbm.rgd; 2618 if (!rgd || !rgrp_contains_block(rgd, block)) 2619 rgd = gfs2_blk2rgrpd(sdp, block, 1); 2620 } 2621 2622 if (!rgd) { 2623 fs_err(sdp, "rlist_add: no rgrp for block %llu\n", 2624 (unsigned long long)block); 2625 return; 2626 } 2627 2628 for (x = 0; x < rlist->rl_rgrps; x++) { 2629 if (rlist->rl_rgd[x] == rgd) { 2630 swap(rlist->rl_rgd[x], 2631 rlist->rl_rgd[rlist->rl_rgrps - 1]); 2632 return; 2633 } 2634 } 2635 2636 if (rlist->rl_rgrps == rlist->rl_space) { 2637 new_space = rlist->rl_space + 10; 2638 2639 tmp = kcalloc(new_space, sizeof(struct gfs2_rgrpd *), 2640 GFP_NOFS | __GFP_NOFAIL); 2641 2642 if (rlist->rl_rgd) { 2643 memcpy(tmp, rlist->rl_rgd, 2644 rlist->rl_space * sizeof(struct gfs2_rgrpd *)); 2645 kfree(rlist->rl_rgd); 2646 } 2647 2648 rlist->rl_space = new_space; 2649 rlist->rl_rgd = tmp; 2650 } 2651 2652 rlist->rl_rgd[rlist->rl_rgrps++] = rgd; 2653 } 2654 2655 /** 2656 * gfs2_rlist_alloc - all RGs have been added to the rlist, now allocate 2657 * and initialize an array of glock holders for them 2658 * @rlist: the list of resource groups 2659 * 2660 * FIXME: Don't use NOFAIL 2661 * 2662 */ 2663 2664 void gfs2_rlist_alloc(struct gfs2_rgrp_list *rlist) 2665 { 2666 unsigned int x; 2667 2668 rlist->rl_ghs = kmalloc_array(rlist->rl_rgrps, 2669 sizeof(struct gfs2_holder), 2670 GFP_NOFS | __GFP_NOFAIL); 2671 for (x = 0; x < rlist->rl_rgrps; x++) 2672 gfs2_holder_init(rlist->rl_rgd[x]->rd_gl, 2673 LM_ST_EXCLUSIVE, 0, 2674 &rlist->rl_ghs[x]); 2675 } 2676 2677 /** 2678 * gfs2_rlist_free - free a resource group list 2679 * @rlist: the list of resource groups 2680 * 2681 */ 2682 2683 void gfs2_rlist_free(struct gfs2_rgrp_list *rlist) 2684 { 2685 unsigned int x; 2686 2687 kfree(rlist->rl_rgd); 2688 2689 if (rlist->rl_ghs) { 2690 for (x = 0; x < rlist->rl_rgrps; x++) 2691 gfs2_holder_uninit(&rlist->rl_ghs[x]); 2692 kfree(rlist->rl_ghs); 2693 rlist->rl_ghs = NULL; 2694 } 2695 } 2696 2697