1 // SPDX-License-Identifier: GPL-2.0-only 2 /* 3 * Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved. 4 * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved. 5 */ 6 7 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt 8 9 #include <linux/slab.h> 10 #include <linux/spinlock.h> 11 #include <linux/completion.h> 12 #include <linux/buffer_head.h> 13 #include <linux/fs.h> 14 #include <linux/gfs2_ondisk.h> 15 #include <linux/prefetch.h> 16 #include <linux/blkdev.h> 17 #include <linux/rbtree.h> 18 #include <linux/random.h> 19 20 #include "gfs2.h" 21 #include "incore.h" 22 #include "glock.h" 23 #include "glops.h" 24 #include "lops.h" 25 #include "meta_io.h" 26 #include "quota.h" 27 #include "rgrp.h" 28 #include "super.h" 29 #include "trans.h" 30 #include "util.h" 31 #include "log.h" 32 #include "inode.h" 33 #include "trace_gfs2.h" 34 #include "dir.h" 35 36 #define BFITNOENT ((u32)~0) 37 #define NO_BLOCK ((u64)~0) 38 39 /* 40 * These routines are used by the resource group routines (rgrp.c) 41 * to keep track of block allocation. Each block is represented by two 42 * bits. So, each byte represents GFS2_NBBY (i.e. 4) blocks. 43 * 44 * 0 = Free 45 * 1 = Used (not metadata) 46 * 2 = Unlinked (still in use) inode 47 * 3 = Used (metadata) 48 */ 49 50 struct gfs2_extent { 51 struct gfs2_rbm rbm; 52 u32 len; 53 }; 54 55 static const char valid_change[16] = { 56 /* current */ 57 /* n */ 0, 1, 1, 1, 58 /* e */ 1, 0, 0, 0, 59 /* w */ 0, 0, 0, 1, 60 1, 0, 0, 0 61 }; 62 63 static int gfs2_rbm_find(struct gfs2_rbm *rbm, u8 state, u32 *minext, 64 const struct gfs2_inode *ip, bool nowrap); 65 66 67 /** 68 * gfs2_setbit - Set a bit in the bitmaps 69 * @rbm: The position of the bit to set 70 * @do_clone: Also set the clone bitmap, if it exists 71 * @new_state: the new state of the block 72 * 73 */ 74 75 static inline void gfs2_setbit(const struct gfs2_rbm *rbm, bool do_clone, 76 unsigned char new_state) 77 { 78 unsigned char *byte1, *byte2, *end, cur_state; 79 struct gfs2_bitmap *bi = rbm_bi(rbm); 80 unsigned int buflen = bi->bi_bytes; 81 const unsigned int bit = (rbm->offset % GFS2_NBBY) * GFS2_BIT_SIZE; 82 83 byte1 = bi->bi_bh->b_data + bi->bi_offset + (rbm->offset / GFS2_NBBY); 84 end = bi->bi_bh->b_data + bi->bi_offset + buflen; 85 86 BUG_ON(byte1 >= end); 87 88 cur_state = (*byte1 >> bit) & GFS2_BIT_MASK; 89 90 if (unlikely(!valid_change[new_state * 4 + cur_state])) { 91 struct gfs2_sbd *sdp = rbm->rgd->rd_sbd; 92 93 fs_warn(sdp, "buf_blk = 0x%x old_state=%d, new_state=%d\n", 94 rbm->offset, cur_state, new_state); 95 fs_warn(sdp, "rgrp=0x%llx bi_start=0x%x biblk: 0x%llx\n", 96 (unsigned long long)rbm->rgd->rd_addr, bi->bi_start, 97 (unsigned long long)bi->bi_bh->b_blocknr); 98 fs_warn(sdp, "bi_offset=0x%x bi_bytes=0x%x block=0x%llx\n", 99 bi->bi_offset, bi->bi_bytes, 100 (unsigned long long)gfs2_rbm_to_block(rbm)); 101 dump_stack(); 102 gfs2_consist_rgrpd(rbm->rgd); 103 return; 104 } 105 *byte1 ^= (cur_state ^ new_state) << bit; 106 107 if (do_clone && bi->bi_clone) { 108 byte2 = bi->bi_clone + bi->bi_offset + (rbm->offset / GFS2_NBBY); 109 cur_state = (*byte2 >> bit) & GFS2_BIT_MASK; 110 *byte2 ^= (cur_state ^ new_state) << bit; 111 } 112 } 113 114 /** 115 * gfs2_testbit - test a bit in the bitmaps 116 * @rbm: The bit to test 117 * @use_clone: If true, test the clone bitmap, not the official bitmap. 118 * 119 * Some callers like gfs2_unaligned_extlen need to test the clone bitmaps, 120 * not the "real" bitmaps, to avoid allocating recently freed blocks. 121 * 122 * Returns: The two bit block state of the requested bit 123 */ 124 125 static inline u8 gfs2_testbit(const struct gfs2_rbm *rbm, bool use_clone) 126 { 127 struct gfs2_bitmap *bi = rbm_bi(rbm); 128 const u8 *buffer; 129 const u8 *byte; 130 unsigned int bit; 131 132 if (use_clone && bi->bi_clone) 133 buffer = bi->bi_clone; 134 else 135 buffer = bi->bi_bh->b_data; 136 buffer += bi->bi_offset; 137 byte = buffer + (rbm->offset / GFS2_NBBY); 138 bit = (rbm->offset % GFS2_NBBY) * GFS2_BIT_SIZE; 139 140 return (*byte >> bit) & GFS2_BIT_MASK; 141 } 142 143 /** 144 * gfs2_bit_search 145 * @ptr: Pointer to bitmap data 146 * @mask: Mask to use (normally 0x55555.... but adjusted for search start) 147 * @state: The state we are searching for 148 * 149 * We xor the bitmap data with a patter which is the bitwise opposite 150 * of what we are looking for, this gives rise to a pattern of ones 151 * wherever there is a match. Since we have two bits per entry, we 152 * take this pattern, shift it down by one place and then and it with 153 * the original. All the even bit positions (0,2,4, etc) then represent 154 * successful matches, so we mask with 0x55555..... to remove the unwanted 155 * odd bit positions. 156 * 157 * This allows searching of a whole u64 at once (32 blocks) with a 158 * single test (on 64 bit arches). 159 */ 160 161 static inline u64 gfs2_bit_search(const __le64 *ptr, u64 mask, u8 state) 162 { 163 u64 tmp; 164 static const u64 search[] = { 165 [0] = 0xffffffffffffffffULL, 166 [1] = 0xaaaaaaaaaaaaaaaaULL, 167 [2] = 0x5555555555555555ULL, 168 [3] = 0x0000000000000000ULL, 169 }; 170 tmp = le64_to_cpu(*ptr) ^ search[state]; 171 tmp &= (tmp >> 1); 172 tmp &= mask; 173 return tmp; 174 } 175 176 /** 177 * rs_cmp - multi-block reservation range compare 178 * @blk: absolute file system block number of the new reservation 179 * @len: number of blocks in the new reservation 180 * @rs: existing reservation to compare against 181 * 182 * returns: 1 if the block range is beyond the reach of the reservation 183 * -1 if the block range is before the start of the reservation 184 * 0 if the block range overlaps with the reservation 185 */ 186 static inline int rs_cmp(u64 blk, u32 len, struct gfs2_blkreserv *rs) 187 { 188 u64 startblk = gfs2_rbm_to_block(&rs->rs_rbm); 189 190 if (blk >= startblk + rs->rs_free) 191 return 1; 192 if (blk + len - 1 < startblk) 193 return -1; 194 return 0; 195 } 196 197 /** 198 * gfs2_bitfit - Search an rgrp's bitmap buffer to find a bit-pair representing 199 * a block in a given allocation state. 200 * @buf: the buffer that holds the bitmaps 201 * @len: the length (in bytes) of the buffer 202 * @goal: start search at this block's bit-pair (within @buffer) 203 * @state: GFS2_BLKST_XXX the state of the block we're looking for. 204 * 205 * Scope of @goal and returned block number is only within this bitmap buffer, 206 * not entire rgrp or filesystem. @buffer will be offset from the actual 207 * beginning of a bitmap block buffer, skipping any header structures, but 208 * headers are always a multiple of 64 bits long so that the buffer is 209 * always aligned to a 64 bit boundary. 210 * 211 * The size of the buffer is in bytes, but is it assumed that it is 212 * always ok to read a complete multiple of 64 bits at the end 213 * of the block in case the end is no aligned to a natural boundary. 214 * 215 * Return: the block number (bitmap buffer scope) that was found 216 */ 217 218 static u32 gfs2_bitfit(const u8 *buf, const unsigned int len, 219 u32 goal, u8 state) 220 { 221 u32 spoint = (goal << 1) & ((8*sizeof(u64)) - 1); 222 const __le64 *ptr = ((__le64 *)buf) + (goal >> 5); 223 const __le64 *end = (__le64 *)(buf + ALIGN(len, sizeof(u64))); 224 u64 tmp; 225 u64 mask = 0x5555555555555555ULL; 226 u32 bit; 227 228 /* Mask off bits we don't care about at the start of the search */ 229 mask <<= spoint; 230 tmp = gfs2_bit_search(ptr, mask, state); 231 ptr++; 232 while(tmp == 0 && ptr < end) { 233 tmp = gfs2_bit_search(ptr, 0x5555555555555555ULL, state); 234 ptr++; 235 } 236 /* Mask off any bits which are more than len bytes from the start */ 237 if (ptr == end && (len & (sizeof(u64) - 1))) 238 tmp &= (((u64)~0) >> (64 - 8*(len & (sizeof(u64) - 1)))); 239 /* Didn't find anything, so return */ 240 if (tmp == 0) 241 return BFITNOENT; 242 ptr--; 243 bit = __ffs64(tmp); 244 bit /= 2; /* two bits per entry in the bitmap */ 245 return (((const unsigned char *)ptr - buf) * GFS2_NBBY) + bit; 246 } 247 248 /** 249 * gfs2_rbm_from_block - Set the rbm based upon rgd and block number 250 * @rbm: The rbm with rgd already set correctly 251 * @block: The block number (filesystem relative) 252 * 253 * This sets the bi and offset members of an rbm based on a 254 * resource group and a filesystem relative block number. The 255 * resource group must be set in the rbm on entry, the bi and 256 * offset members will be set by this function. 257 * 258 * Returns: 0 on success, or an error code 259 */ 260 261 static int gfs2_rbm_from_block(struct gfs2_rbm *rbm, u64 block) 262 { 263 if (!rgrp_contains_block(rbm->rgd, block)) 264 return -E2BIG; 265 rbm->bii = 0; 266 rbm->offset = block - rbm->rgd->rd_data0; 267 /* Check if the block is within the first block */ 268 if (rbm->offset < rbm_bi(rbm)->bi_blocks) 269 return 0; 270 271 /* Adjust for the size diff between gfs2_meta_header and gfs2_rgrp */ 272 rbm->offset += (sizeof(struct gfs2_rgrp) - 273 sizeof(struct gfs2_meta_header)) * GFS2_NBBY; 274 rbm->bii = rbm->offset / rbm->rgd->rd_sbd->sd_blocks_per_bitmap; 275 rbm->offset -= rbm->bii * rbm->rgd->rd_sbd->sd_blocks_per_bitmap; 276 return 0; 277 } 278 279 /** 280 * gfs2_rbm_incr - increment an rbm structure 281 * @rbm: The rbm with rgd already set correctly 282 * 283 * This function takes an existing rbm structure and increments it to the next 284 * viable block offset. 285 * 286 * Returns: If incrementing the offset would cause the rbm to go past the 287 * end of the rgrp, true is returned, otherwise false. 288 * 289 */ 290 291 static bool gfs2_rbm_incr(struct gfs2_rbm *rbm) 292 { 293 if (rbm->offset + 1 < rbm_bi(rbm)->bi_blocks) { /* in the same bitmap */ 294 rbm->offset++; 295 return false; 296 } 297 if (rbm->bii == rbm->rgd->rd_length - 1) /* at the last bitmap */ 298 return true; 299 300 rbm->offset = 0; 301 rbm->bii++; 302 return false; 303 } 304 305 /** 306 * gfs2_unaligned_extlen - Look for free blocks which are not byte aligned 307 * @rbm: Position to search (value/result) 308 * @n_unaligned: Number of unaligned blocks to check 309 * @len: Decremented for each block found (terminate on zero) 310 * 311 * Returns: true if a non-free block is encountered 312 */ 313 314 static bool gfs2_unaligned_extlen(struct gfs2_rbm *rbm, u32 n_unaligned, u32 *len) 315 { 316 u32 n; 317 u8 res; 318 319 for (n = 0; n < n_unaligned; n++) { 320 res = gfs2_testbit(rbm, true); 321 if (res != GFS2_BLKST_FREE) 322 return true; 323 (*len)--; 324 if (*len == 0) 325 return true; 326 if (gfs2_rbm_incr(rbm)) 327 return true; 328 } 329 330 return false; 331 } 332 333 /** 334 * gfs2_free_extlen - Return extent length of free blocks 335 * @rrbm: Starting position 336 * @len: Max length to check 337 * 338 * Starting at the block specified by the rbm, see how many free blocks 339 * there are, not reading more than len blocks ahead. This can be done 340 * using memchr_inv when the blocks are byte aligned, but has to be done 341 * on a block by block basis in case of unaligned blocks. Also this 342 * function can cope with bitmap boundaries (although it must stop on 343 * a resource group boundary) 344 * 345 * Returns: Number of free blocks in the extent 346 */ 347 348 static u32 gfs2_free_extlen(const struct gfs2_rbm *rrbm, u32 len) 349 { 350 struct gfs2_rbm rbm = *rrbm; 351 u32 n_unaligned = rbm.offset & 3; 352 u32 size = len; 353 u32 bytes; 354 u32 chunk_size; 355 u8 *ptr, *start, *end; 356 u64 block; 357 struct gfs2_bitmap *bi; 358 359 if (n_unaligned && 360 gfs2_unaligned_extlen(&rbm, 4 - n_unaligned, &len)) 361 goto out; 362 363 n_unaligned = len & 3; 364 /* Start is now byte aligned */ 365 while (len > 3) { 366 bi = rbm_bi(&rbm); 367 start = bi->bi_bh->b_data; 368 if (bi->bi_clone) 369 start = bi->bi_clone; 370 start += bi->bi_offset; 371 end = start + bi->bi_bytes; 372 BUG_ON(rbm.offset & 3); 373 start += (rbm.offset / GFS2_NBBY); 374 bytes = min_t(u32, len / GFS2_NBBY, (end - start)); 375 ptr = memchr_inv(start, 0, bytes); 376 chunk_size = ((ptr == NULL) ? bytes : (ptr - start)); 377 chunk_size *= GFS2_NBBY; 378 BUG_ON(len < chunk_size); 379 len -= chunk_size; 380 block = gfs2_rbm_to_block(&rbm); 381 if (gfs2_rbm_from_block(&rbm, block + chunk_size)) { 382 n_unaligned = 0; 383 break; 384 } 385 if (ptr) { 386 n_unaligned = 3; 387 break; 388 } 389 n_unaligned = len & 3; 390 } 391 392 /* Deal with any bits left over at the end */ 393 if (n_unaligned) 394 gfs2_unaligned_extlen(&rbm, n_unaligned, &len); 395 out: 396 return size - len; 397 } 398 399 /** 400 * gfs2_bitcount - count the number of bits in a certain state 401 * @rgd: the resource group descriptor 402 * @buffer: the buffer that holds the bitmaps 403 * @buflen: the length (in bytes) of the buffer 404 * @state: the state of the block we're looking for 405 * 406 * Returns: The number of bits 407 */ 408 409 static u32 gfs2_bitcount(struct gfs2_rgrpd *rgd, const u8 *buffer, 410 unsigned int buflen, u8 state) 411 { 412 const u8 *byte = buffer; 413 const u8 *end = buffer + buflen; 414 const u8 state1 = state << 2; 415 const u8 state2 = state << 4; 416 const u8 state3 = state << 6; 417 u32 count = 0; 418 419 for (; byte < end; byte++) { 420 if (((*byte) & 0x03) == state) 421 count++; 422 if (((*byte) & 0x0C) == state1) 423 count++; 424 if (((*byte) & 0x30) == state2) 425 count++; 426 if (((*byte) & 0xC0) == state3) 427 count++; 428 } 429 430 return count; 431 } 432 433 /** 434 * gfs2_rgrp_verify - Verify that a resource group is consistent 435 * @rgd: the rgrp 436 * 437 */ 438 439 void gfs2_rgrp_verify(struct gfs2_rgrpd *rgd) 440 { 441 struct gfs2_sbd *sdp = rgd->rd_sbd; 442 struct gfs2_bitmap *bi = NULL; 443 u32 length = rgd->rd_length; 444 u32 count[4], tmp; 445 int buf, x; 446 447 memset(count, 0, 4 * sizeof(u32)); 448 449 /* Count # blocks in each of 4 possible allocation states */ 450 for (buf = 0; buf < length; buf++) { 451 bi = rgd->rd_bits + buf; 452 for (x = 0; x < 4; x++) 453 count[x] += gfs2_bitcount(rgd, 454 bi->bi_bh->b_data + 455 bi->bi_offset, 456 bi->bi_bytes, x); 457 } 458 459 if (count[0] != rgd->rd_free) { 460 if (gfs2_consist_rgrpd(rgd)) 461 fs_err(sdp, "free data mismatch: %u != %u\n", 462 count[0], rgd->rd_free); 463 return; 464 } 465 466 tmp = rgd->rd_data - rgd->rd_free - rgd->rd_dinodes; 467 if (count[1] != tmp) { 468 if (gfs2_consist_rgrpd(rgd)) 469 fs_err(sdp, "used data mismatch: %u != %u\n", 470 count[1], tmp); 471 return; 472 } 473 474 if (count[2] + count[3] != rgd->rd_dinodes) { 475 if (gfs2_consist_rgrpd(rgd)) 476 fs_err(sdp, "used metadata mismatch: %u != %u\n", 477 count[2] + count[3], rgd->rd_dinodes); 478 return; 479 } 480 } 481 482 /** 483 * gfs2_blk2rgrpd - Find resource group for a given data/meta block number 484 * @sdp: The GFS2 superblock 485 * @blk: The data block number 486 * @exact: True if this needs to be an exact match 487 * 488 * The @exact argument should be set to true by most callers. The exception 489 * is when we need to match blocks which are not represented by the rgrp 490 * bitmap, but which are part of the rgrp (i.e. padding blocks) which are 491 * there for alignment purposes. Another way of looking at it is that @exact 492 * matches only valid data/metadata blocks, but with @exact false, it will 493 * match any block within the extent of the rgrp. 494 * 495 * Returns: The resource group, or NULL if not found 496 */ 497 498 struct gfs2_rgrpd *gfs2_blk2rgrpd(struct gfs2_sbd *sdp, u64 blk, bool exact) 499 { 500 struct rb_node *n, *next; 501 struct gfs2_rgrpd *cur; 502 503 spin_lock(&sdp->sd_rindex_spin); 504 n = sdp->sd_rindex_tree.rb_node; 505 while (n) { 506 cur = rb_entry(n, struct gfs2_rgrpd, rd_node); 507 next = NULL; 508 if (blk < cur->rd_addr) 509 next = n->rb_left; 510 else if (blk >= cur->rd_data0 + cur->rd_data) 511 next = n->rb_right; 512 if (next == NULL) { 513 spin_unlock(&sdp->sd_rindex_spin); 514 if (exact) { 515 if (blk < cur->rd_addr) 516 return NULL; 517 if (blk >= cur->rd_data0 + cur->rd_data) 518 return NULL; 519 } 520 return cur; 521 } 522 n = next; 523 } 524 spin_unlock(&sdp->sd_rindex_spin); 525 526 return NULL; 527 } 528 529 /** 530 * gfs2_rgrpd_get_first - get the first Resource Group in the filesystem 531 * @sdp: The GFS2 superblock 532 * 533 * Returns: The first rgrp in the filesystem 534 */ 535 536 struct gfs2_rgrpd *gfs2_rgrpd_get_first(struct gfs2_sbd *sdp) 537 { 538 const struct rb_node *n; 539 struct gfs2_rgrpd *rgd; 540 541 spin_lock(&sdp->sd_rindex_spin); 542 n = rb_first(&sdp->sd_rindex_tree); 543 rgd = rb_entry(n, struct gfs2_rgrpd, rd_node); 544 spin_unlock(&sdp->sd_rindex_spin); 545 546 return rgd; 547 } 548 549 /** 550 * gfs2_rgrpd_get_next - get the next RG 551 * @rgd: the resource group descriptor 552 * 553 * Returns: The next rgrp 554 */ 555 556 struct gfs2_rgrpd *gfs2_rgrpd_get_next(struct gfs2_rgrpd *rgd) 557 { 558 struct gfs2_sbd *sdp = rgd->rd_sbd; 559 const struct rb_node *n; 560 561 spin_lock(&sdp->sd_rindex_spin); 562 n = rb_next(&rgd->rd_node); 563 if (n == NULL) 564 n = rb_first(&sdp->sd_rindex_tree); 565 566 if (unlikely(&rgd->rd_node == n)) { 567 spin_unlock(&sdp->sd_rindex_spin); 568 return NULL; 569 } 570 rgd = rb_entry(n, struct gfs2_rgrpd, rd_node); 571 spin_unlock(&sdp->sd_rindex_spin); 572 return rgd; 573 } 574 575 void check_and_update_goal(struct gfs2_inode *ip) 576 { 577 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 578 if (!ip->i_goal || gfs2_blk2rgrpd(sdp, ip->i_goal, 1) == NULL) 579 ip->i_goal = ip->i_no_addr; 580 } 581 582 void gfs2_free_clones(struct gfs2_rgrpd *rgd) 583 { 584 int x; 585 586 for (x = 0; x < rgd->rd_length; x++) { 587 struct gfs2_bitmap *bi = rgd->rd_bits + x; 588 kfree(bi->bi_clone); 589 bi->bi_clone = NULL; 590 } 591 } 592 593 /** 594 * gfs2_rsqa_alloc - make sure we have a reservation assigned to the inode 595 * plus a quota allocations data structure, if necessary 596 * @ip: the inode for this reservation 597 */ 598 int gfs2_rsqa_alloc(struct gfs2_inode *ip) 599 { 600 return gfs2_qa_alloc(ip); 601 } 602 603 static void dump_rs(struct seq_file *seq, const struct gfs2_blkreserv *rs, 604 const char *fs_id_buf) 605 { 606 struct gfs2_inode *ip = container_of(rs, struct gfs2_inode, i_res); 607 608 gfs2_print_dbg(seq, "%s B: n:%llu s:%llu b:%u f:%u\n", fs_id_buf, 609 (unsigned long long)ip->i_no_addr, 610 (unsigned long long)gfs2_rbm_to_block(&rs->rs_rbm), 611 rs->rs_rbm.offset, rs->rs_free); 612 } 613 614 /** 615 * __rs_deltree - remove a multi-block reservation from the rgd tree 616 * @rs: The reservation to remove 617 * 618 */ 619 static void __rs_deltree(struct gfs2_blkreserv *rs) 620 { 621 struct gfs2_rgrpd *rgd; 622 623 if (!gfs2_rs_active(rs)) 624 return; 625 626 rgd = rs->rs_rbm.rgd; 627 trace_gfs2_rs(rs, TRACE_RS_TREEDEL); 628 rb_erase(&rs->rs_node, &rgd->rd_rstree); 629 RB_CLEAR_NODE(&rs->rs_node); 630 631 if (rs->rs_free) { 632 u64 last_block = gfs2_rbm_to_block(&rs->rs_rbm) + 633 rs->rs_free - 1; 634 struct gfs2_rbm last_rbm = { .rgd = rs->rs_rbm.rgd, }; 635 struct gfs2_bitmap *start, *last; 636 637 /* return reserved blocks to the rgrp */ 638 BUG_ON(rs->rs_rbm.rgd->rd_reserved < rs->rs_free); 639 rs->rs_rbm.rgd->rd_reserved -= rs->rs_free; 640 /* The rgrp extent failure point is likely not to increase; 641 it will only do so if the freed blocks are somehow 642 contiguous with a span of free blocks that follows. Still, 643 it will force the number to be recalculated later. */ 644 rgd->rd_extfail_pt += rs->rs_free; 645 rs->rs_free = 0; 646 if (gfs2_rbm_from_block(&last_rbm, last_block)) 647 return; 648 start = rbm_bi(&rs->rs_rbm); 649 last = rbm_bi(&last_rbm); 650 do 651 clear_bit(GBF_FULL, &start->bi_flags); 652 while (start++ != last); 653 } 654 } 655 656 /** 657 * gfs2_rs_deltree - remove a multi-block reservation from the rgd tree 658 * @rs: The reservation to remove 659 * 660 */ 661 void gfs2_rs_deltree(struct gfs2_blkreserv *rs) 662 { 663 struct gfs2_rgrpd *rgd; 664 665 rgd = rs->rs_rbm.rgd; 666 if (rgd) { 667 spin_lock(&rgd->rd_rsspin); 668 __rs_deltree(rs); 669 BUG_ON(rs->rs_free); 670 spin_unlock(&rgd->rd_rsspin); 671 } 672 } 673 674 /** 675 * gfs2_rsqa_delete - delete a multi-block reservation and quota allocation 676 * @ip: The inode for this reservation 677 * @wcount: The inode's write count, or NULL 678 * 679 */ 680 void gfs2_rsqa_delete(struct gfs2_inode *ip, atomic_t *wcount) 681 { 682 down_write(&ip->i_rw_mutex); 683 if ((wcount == NULL) || (atomic_read(wcount) <= 1)) 684 gfs2_rs_deltree(&ip->i_res); 685 up_write(&ip->i_rw_mutex); 686 gfs2_qa_delete(ip, wcount); 687 } 688 689 /** 690 * return_all_reservations - return all reserved blocks back to the rgrp. 691 * @rgd: the rgrp that needs its space back 692 * 693 * We previously reserved a bunch of blocks for allocation. Now we need to 694 * give them back. This leave the reservation structures in tact, but removes 695 * all of their corresponding "no-fly zones". 696 */ 697 static void return_all_reservations(struct gfs2_rgrpd *rgd) 698 { 699 struct rb_node *n; 700 struct gfs2_blkreserv *rs; 701 702 spin_lock(&rgd->rd_rsspin); 703 while ((n = rb_first(&rgd->rd_rstree))) { 704 rs = rb_entry(n, struct gfs2_blkreserv, rs_node); 705 __rs_deltree(rs); 706 } 707 spin_unlock(&rgd->rd_rsspin); 708 } 709 710 void gfs2_clear_rgrpd(struct gfs2_sbd *sdp) 711 { 712 struct rb_node *n; 713 struct gfs2_rgrpd *rgd; 714 struct gfs2_glock *gl; 715 716 while ((n = rb_first(&sdp->sd_rindex_tree))) { 717 rgd = rb_entry(n, struct gfs2_rgrpd, rd_node); 718 gl = rgd->rd_gl; 719 720 rb_erase(n, &sdp->sd_rindex_tree); 721 722 if (gl) { 723 glock_clear_object(gl, rgd); 724 gfs2_rgrp_brelse(rgd); 725 gfs2_glock_put(gl); 726 } 727 728 gfs2_free_clones(rgd); 729 kfree(rgd->rd_bits); 730 rgd->rd_bits = NULL; 731 return_all_reservations(rgd); 732 kmem_cache_free(gfs2_rgrpd_cachep, rgd); 733 } 734 } 735 736 static void gfs2_rindex_print(const struct gfs2_rgrpd *rgd) 737 { 738 struct gfs2_sbd *sdp = rgd->rd_sbd; 739 740 fs_info(sdp, "ri_addr = %llu\n", (unsigned long long)rgd->rd_addr); 741 fs_info(sdp, "ri_length = %u\n", rgd->rd_length); 742 fs_info(sdp, "ri_data0 = %llu\n", (unsigned long long)rgd->rd_data0); 743 fs_info(sdp, "ri_data = %u\n", rgd->rd_data); 744 fs_info(sdp, "ri_bitbytes = %u\n", rgd->rd_bitbytes); 745 } 746 747 /** 748 * gfs2_compute_bitstructs - Compute the bitmap sizes 749 * @rgd: The resource group descriptor 750 * 751 * Calculates bitmap descriptors, one for each block that contains bitmap data 752 * 753 * Returns: errno 754 */ 755 756 static int compute_bitstructs(struct gfs2_rgrpd *rgd) 757 { 758 struct gfs2_sbd *sdp = rgd->rd_sbd; 759 struct gfs2_bitmap *bi; 760 u32 length = rgd->rd_length; /* # blocks in hdr & bitmap */ 761 u32 bytes_left, bytes; 762 int x; 763 764 if (!length) 765 return -EINVAL; 766 767 rgd->rd_bits = kcalloc(length, sizeof(struct gfs2_bitmap), GFP_NOFS); 768 if (!rgd->rd_bits) 769 return -ENOMEM; 770 771 bytes_left = rgd->rd_bitbytes; 772 773 for (x = 0; x < length; x++) { 774 bi = rgd->rd_bits + x; 775 776 bi->bi_flags = 0; 777 /* small rgrp; bitmap stored completely in header block */ 778 if (length == 1) { 779 bytes = bytes_left; 780 bi->bi_offset = sizeof(struct gfs2_rgrp); 781 bi->bi_start = 0; 782 bi->bi_bytes = bytes; 783 bi->bi_blocks = bytes * GFS2_NBBY; 784 /* header block */ 785 } else if (x == 0) { 786 bytes = sdp->sd_sb.sb_bsize - sizeof(struct gfs2_rgrp); 787 bi->bi_offset = sizeof(struct gfs2_rgrp); 788 bi->bi_start = 0; 789 bi->bi_bytes = bytes; 790 bi->bi_blocks = bytes * GFS2_NBBY; 791 /* last block */ 792 } else if (x + 1 == length) { 793 bytes = bytes_left; 794 bi->bi_offset = sizeof(struct gfs2_meta_header); 795 bi->bi_start = rgd->rd_bitbytes - bytes_left; 796 bi->bi_bytes = bytes; 797 bi->bi_blocks = bytes * GFS2_NBBY; 798 /* other blocks */ 799 } else { 800 bytes = sdp->sd_sb.sb_bsize - 801 sizeof(struct gfs2_meta_header); 802 bi->bi_offset = sizeof(struct gfs2_meta_header); 803 bi->bi_start = rgd->rd_bitbytes - bytes_left; 804 bi->bi_bytes = bytes; 805 bi->bi_blocks = bytes * GFS2_NBBY; 806 } 807 808 bytes_left -= bytes; 809 } 810 811 if (bytes_left) { 812 gfs2_consist_rgrpd(rgd); 813 return -EIO; 814 } 815 bi = rgd->rd_bits + (length - 1); 816 if ((bi->bi_start + bi->bi_bytes) * GFS2_NBBY != rgd->rd_data) { 817 if (gfs2_consist_rgrpd(rgd)) { 818 gfs2_rindex_print(rgd); 819 fs_err(sdp, "start=%u len=%u offset=%u\n", 820 bi->bi_start, bi->bi_bytes, bi->bi_offset); 821 } 822 return -EIO; 823 } 824 825 return 0; 826 } 827 828 /** 829 * gfs2_ri_total - Total up the file system space, according to the rindex. 830 * @sdp: the filesystem 831 * 832 */ 833 u64 gfs2_ri_total(struct gfs2_sbd *sdp) 834 { 835 u64 total_data = 0; 836 struct inode *inode = sdp->sd_rindex; 837 struct gfs2_inode *ip = GFS2_I(inode); 838 char buf[sizeof(struct gfs2_rindex)]; 839 int error, rgrps; 840 841 for (rgrps = 0;; rgrps++) { 842 loff_t pos = rgrps * sizeof(struct gfs2_rindex); 843 844 if (pos + sizeof(struct gfs2_rindex) > i_size_read(inode)) 845 break; 846 error = gfs2_internal_read(ip, buf, &pos, 847 sizeof(struct gfs2_rindex)); 848 if (error != sizeof(struct gfs2_rindex)) 849 break; 850 total_data += be32_to_cpu(((struct gfs2_rindex *)buf)->ri_data); 851 } 852 return total_data; 853 } 854 855 static int rgd_insert(struct gfs2_rgrpd *rgd) 856 { 857 struct gfs2_sbd *sdp = rgd->rd_sbd; 858 struct rb_node **newn = &sdp->sd_rindex_tree.rb_node, *parent = NULL; 859 860 /* Figure out where to put new node */ 861 while (*newn) { 862 struct gfs2_rgrpd *cur = rb_entry(*newn, struct gfs2_rgrpd, 863 rd_node); 864 865 parent = *newn; 866 if (rgd->rd_addr < cur->rd_addr) 867 newn = &((*newn)->rb_left); 868 else if (rgd->rd_addr > cur->rd_addr) 869 newn = &((*newn)->rb_right); 870 else 871 return -EEXIST; 872 } 873 874 rb_link_node(&rgd->rd_node, parent, newn); 875 rb_insert_color(&rgd->rd_node, &sdp->sd_rindex_tree); 876 sdp->sd_rgrps++; 877 return 0; 878 } 879 880 /** 881 * read_rindex_entry - Pull in a new resource index entry from the disk 882 * @ip: Pointer to the rindex inode 883 * 884 * Returns: 0 on success, > 0 on EOF, error code otherwise 885 */ 886 887 static int read_rindex_entry(struct gfs2_inode *ip) 888 { 889 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 890 const unsigned bsize = sdp->sd_sb.sb_bsize; 891 loff_t pos = sdp->sd_rgrps * sizeof(struct gfs2_rindex); 892 struct gfs2_rindex buf; 893 int error; 894 struct gfs2_rgrpd *rgd; 895 896 if (pos >= i_size_read(&ip->i_inode)) 897 return 1; 898 899 error = gfs2_internal_read(ip, (char *)&buf, &pos, 900 sizeof(struct gfs2_rindex)); 901 902 if (error != sizeof(struct gfs2_rindex)) 903 return (error == 0) ? 1 : error; 904 905 rgd = kmem_cache_zalloc(gfs2_rgrpd_cachep, GFP_NOFS); 906 error = -ENOMEM; 907 if (!rgd) 908 return error; 909 910 rgd->rd_sbd = sdp; 911 rgd->rd_addr = be64_to_cpu(buf.ri_addr); 912 rgd->rd_length = be32_to_cpu(buf.ri_length); 913 rgd->rd_data0 = be64_to_cpu(buf.ri_data0); 914 rgd->rd_data = be32_to_cpu(buf.ri_data); 915 rgd->rd_bitbytes = be32_to_cpu(buf.ri_bitbytes); 916 spin_lock_init(&rgd->rd_rsspin); 917 918 error = compute_bitstructs(rgd); 919 if (error) 920 goto fail; 921 922 error = gfs2_glock_get(sdp, rgd->rd_addr, 923 &gfs2_rgrp_glops, CREATE, &rgd->rd_gl); 924 if (error) 925 goto fail; 926 927 rgd->rd_rgl = (struct gfs2_rgrp_lvb *)rgd->rd_gl->gl_lksb.sb_lvbptr; 928 rgd->rd_flags &= ~(GFS2_RDF_UPTODATE | GFS2_RDF_PREFERRED); 929 if (rgd->rd_data > sdp->sd_max_rg_data) 930 sdp->sd_max_rg_data = rgd->rd_data; 931 spin_lock(&sdp->sd_rindex_spin); 932 error = rgd_insert(rgd); 933 spin_unlock(&sdp->sd_rindex_spin); 934 if (!error) { 935 glock_set_object(rgd->rd_gl, rgd); 936 rgd->rd_gl->gl_vm.start = (rgd->rd_addr * bsize) & PAGE_MASK; 937 rgd->rd_gl->gl_vm.end = PAGE_ALIGN((rgd->rd_addr + 938 rgd->rd_length) * bsize) - 1; 939 return 0; 940 } 941 942 error = 0; /* someone else read in the rgrp; free it and ignore it */ 943 gfs2_glock_put(rgd->rd_gl); 944 945 fail: 946 kfree(rgd->rd_bits); 947 rgd->rd_bits = NULL; 948 kmem_cache_free(gfs2_rgrpd_cachep, rgd); 949 return error; 950 } 951 952 /** 953 * set_rgrp_preferences - Run all the rgrps, selecting some we prefer to use 954 * @sdp: the GFS2 superblock 955 * 956 * The purpose of this function is to select a subset of the resource groups 957 * and mark them as PREFERRED. We do it in such a way that each node prefers 958 * to use a unique set of rgrps to minimize glock contention. 959 */ 960 static void set_rgrp_preferences(struct gfs2_sbd *sdp) 961 { 962 struct gfs2_rgrpd *rgd, *first; 963 int i; 964 965 /* Skip an initial number of rgrps, based on this node's journal ID. 966 That should start each node out on its own set. */ 967 rgd = gfs2_rgrpd_get_first(sdp); 968 for (i = 0; i < sdp->sd_lockstruct.ls_jid; i++) 969 rgd = gfs2_rgrpd_get_next(rgd); 970 first = rgd; 971 972 do { 973 rgd->rd_flags |= GFS2_RDF_PREFERRED; 974 for (i = 0; i < sdp->sd_journals; i++) { 975 rgd = gfs2_rgrpd_get_next(rgd); 976 if (!rgd || rgd == first) 977 break; 978 } 979 } while (rgd && rgd != first); 980 } 981 982 /** 983 * gfs2_ri_update - Pull in a new resource index from the disk 984 * @ip: pointer to the rindex inode 985 * 986 * Returns: 0 on successful update, error code otherwise 987 */ 988 989 static int gfs2_ri_update(struct gfs2_inode *ip) 990 { 991 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 992 int error; 993 994 do { 995 error = read_rindex_entry(ip); 996 } while (error == 0); 997 998 if (error < 0) 999 return error; 1000 1001 set_rgrp_preferences(sdp); 1002 1003 sdp->sd_rindex_uptodate = 1; 1004 return 0; 1005 } 1006 1007 /** 1008 * gfs2_rindex_update - Update the rindex if required 1009 * @sdp: The GFS2 superblock 1010 * 1011 * We grab a lock on the rindex inode to make sure that it doesn't 1012 * change whilst we are performing an operation. We keep this lock 1013 * for quite long periods of time compared to other locks. This 1014 * doesn't matter, since it is shared and it is very, very rarely 1015 * accessed in the exclusive mode (i.e. only when expanding the filesystem). 1016 * 1017 * This makes sure that we're using the latest copy of the resource index 1018 * special file, which might have been updated if someone expanded the 1019 * filesystem (via gfs2_grow utility), which adds new resource groups. 1020 * 1021 * Returns: 0 on succeess, error code otherwise 1022 */ 1023 1024 int gfs2_rindex_update(struct gfs2_sbd *sdp) 1025 { 1026 struct gfs2_inode *ip = GFS2_I(sdp->sd_rindex); 1027 struct gfs2_glock *gl = ip->i_gl; 1028 struct gfs2_holder ri_gh; 1029 int error = 0; 1030 int unlock_required = 0; 1031 1032 /* Read new copy from disk if we don't have the latest */ 1033 if (!sdp->sd_rindex_uptodate) { 1034 if (!gfs2_glock_is_locked_by_me(gl)) { 1035 error = gfs2_glock_nq_init(gl, LM_ST_SHARED, 0, &ri_gh); 1036 if (error) 1037 return error; 1038 unlock_required = 1; 1039 } 1040 if (!sdp->sd_rindex_uptodate) 1041 error = gfs2_ri_update(ip); 1042 if (unlock_required) 1043 gfs2_glock_dq_uninit(&ri_gh); 1044 } 1045 1046 return error; 1047 } 1048 1049 static void gfs2_rgrp_in(struct gfs2_rgrpd *rgd, const void *buf) 1050 { 1051 const struct gfs2_rgrp *str = buf; 1052 u32 rg_flags; 1053 1054 rg_flags = be32_to_cpu(str->rg_flags); 1055 rg_flags &= ~GFS2_RDF_MASK; 1056 rgd->rd_flags &= GFS2_RDF_MASK; 1057 rgd->rd_flags |= rg_flags; 1058 rgd->rd_free = be32_to_cpu(str->rg_free); 1059 rgd->rd_dinodes = be32_to_cpu(str->rg_dinodes); 1060 rgd->rd_igeneration = be64_to_cpu(str->rg_igeneration); 1061 /* rd_data0, rd_data and rd_bitbytes already set from rindex */ 1062 } 1063 1064 static void gfs2_rgrp_ondisk2lvb(struct gfs2_rgrp_lvb *rgl, const void *buf) 1065 { 1066 const struct gfs2_rgrp *str = buf; 1067 1068 rgl->rl_magic = cpu_to_be32(GFS2_MAGIC); 1069 rgl->rl_flags = str->rg_flags; 1070 rgl->rl_free = str->rg_free; 1071 rgl->rl_dinodes = str->rg_dinodes; 1072 rgl->rl_igeneration = str->rg_igeneration; 1073 rgl->__pad = 0UL; 1074 } 1075 1076 static void gfs2_rgrp_out(struct gfs2_rgrpd *rgd, void *buf) 1077 { 1078 struct gfs2_rgrpd *next = gfs2_rgrpd_get_next(rgd); 1079 struct gfs2_rgrp *str = buf; 1080 u32 crc; 1081 1082 str->rg_flags = cpu_to_be32(rgd->rd_flags & ~GFS2_RDF_MASK); 1083 str->rg_free = cpu_to_be32(rgd->rd_free); 1084 str->rg_dinodes = cpu_to_be32(rgd->rd_dinodes); 1085 if (next == NULL) 1086 str->rg_skip = 0; 1087 else if (next->rd_addr > rgd->rd_addr) 1088 str->rg_skip = cpu_to_be32(next->rd_addr - rgd->rd_addr); 1089 str->rg_igeneration = cpu_to_be64(rgd->rd_igeneration); 1090 str->rg_data0 = cpu_to_be64(rgd->rd_data0); 1091 str->rg_data = cpu_to_be32(rgd->rd_data); 1092 str->rg_bitbytes = cpu_to_be32(rgd->rd_bitbytes); 1093 str->rg_crc = 0; 1094 crc = gfs2_disk_hash(buf, sizeof(struct gfs2_rgrp)); 1095 str->rg_crc = cpu_to_be32(crc); 1096 1097 memset(&str->rg_reserved, 0, sizeof(str->rg_reserved)); 1098 gfs2_rgrp_ondisk2lvb(rgd->rd_rgl, buf); 1099 } 1100 1101 static int gfs2_rgrp_lvb_valid(struct gfs2_rgrpd *rgd) 1102 { 1103 struct gfs2_rgrp_lvb *rgl = rgd->rd_rgl; 1104 struct gfs2_rgrp *str = (struct gfs2_rgrp *)rgd->rd_bits[0].bi_bh->b_data; 1105 struct gfs2_sbd *sdp = rgd->rd_sbd; 1106 int valid = 1; 1107 1108 if (rgl->rl_flags != str->rg_flags) { 1109 fs_warn(sdp, "GFS2: rgd: %llu lvb flag mismatch %u/%u", 1110 (unsigned long long)rgd->rd_addr, 1111 be32_to_cpu(rgl->rl_flags), be32_to_cpu(str->rg_flags)); 1112 valid = 0; 1113 } 1114 if (rgl->rl_free != str->rg_free) { 1115 fs_warn(sdp, "GFS2: rgd: %llu lvb free mismatch %u/%u", 1116 (unsigned long long)rgd->rd_addr, 1117 be32_to_cpu(rgl->rl_free), be32_to_cpu(str->rg_free)); 1118 valid = 0; 1119 } 1120 if (rgl->rl_dinodes != str->rg_dinodes) { 1121 fs_warn(sdp, "GFS2: rgd: %llu lvb dinode mismatch %u/%u", 1122 (unsigned long long)rgd->rd_addr, 1123 be32_to_cpu(rgl->rl_dinodes), 1124 be32_to_cpu(str->rg_dinodes)); 1125 valid = 0; 1126 } 1127 if (rgl->rl_igeneration != str->rg_igeneration) { 1128 fs_warn(sdp, "GFS2: rgd: %llu lvb igen mismatch %llu/%llu", 1129 (unsigned long long)rgd->rd_addr, 1130 (unsigned long long)be64_to_cpu(rgl->rl_igeneration), 1131 (unsigned long long)be64_to_cpu(str->rg_igeneration)); 1132 valid = 0; 1133 } 1134 return valid; 1135 } 1136 1137 static u32 count_unlinked(struct gfs2_rgrpd *rgd) 1138 { 1139 struct gfs2_bitmap *bi; 1140 const u32 length = rgd->rd_length; 1141 const u8 *buffer = NULL; 1142 u32 i, goal, count = 0; 1143 1144 for (i = 0, bi = rgd->rd_bits; i < length; i++, bi++) { 1145 goal = 0; 1146 buffer = bi->bi_bh->b_data + bi->bi_offset; 1147 WARN_ON(!buffer_uptodate(bi->bi_bh)); 1148 while (goal < bi->bi_blocks) { 1149 goal = gfs2_bitfit(buffer, bi->bi_bytes, goal, 1150 GFS2_BLKST_UNLINKED); 1151 if (goal == BFITNOENT) 1152 break; 1153 count++; 1154 goal++; 1155 } 1156 } 1157 1158 return count; 1159 } 1160 1161 1162 /** 1163 * gfs2_rgrp_bh_get - Read in a RG's header and bitmaps 1164 * @rgd: the struct gfs2_rgrpd describing the RG to read in 1165 * 1166 * Read in all of a Resource Group's header and bitmap blocks. 1167 * Caller must eventually call gfs2_rgrp_brelse() to free the bitmaps. 1168 * 1169 * Returns: errno 1170 */ 1171 1172 static int gfs2_rgrp_bh_get(struct gfs2_rgrpd *rgd) 1173 { 1174 struct gfs2_sbd *sdp = rgd->rd_sbd; 1175 struct gfs2_glock *gl = rgd->rd_gl; 1176 unsigned int length = rgd->rd_length; 1177 struct gfs2_bitmap *bi; 1178 unsigned int x, y; 1179 int error; 1180 1181 if (rgd->rd_bits[0].bi_bh != NULL) 1182 return 0; 1183 1184 for (x = 0; x < length; x++) { 1185 bi = rgd->rd_bits + x; 1186 error = gfs2_meta_read(gl, rgd->rd_addr + x, 0, 0, &bi->bi_bh); 1187 if (error) 1188 goto fail; 1189 } 1190 1191 for (y = length; y--;) { 1192 bi = rgd->rd_bits + y; 1193 error = gfs2_meta_wait(sdp, bi->bi_bh); 1194 if (error) 1195 goto fail; 1196 if (gfs2_metatype_check(sdp, bi->bi_bh, y ? GFS2_METATYPE_RB : 1197 GFS2_METATYPE_RG)) { 1198 error = -EIO; 1199 goto fail; 1200 } 1201 } 1202 1203 if (!(rgd->rd_flags & GFS2_RDF_UPTODATE)) { 1204 for (x = 0; x < length; x++) 1205 clear_bit(GBF_FULL, &rgd->rd_bits[x].bi_flags); 1206 gfs2_rgrp_in(rgd, (rgd->rd_bits[0].bi_bh)->b_data); 1207 rgd->rd_flags |= (GFS2_RDF_UPTODATE | GFS2_RDF_CHECK); 1208 rgd->rd_free_clone = rgd->rd_free; 1209 /* max out the rgrp allocation failure point */ 1210 rgd->rd_extfail_pt = rgd->rd_free; 1211 } 1212 if (cpu_to_be32(GFS2_MAGIC) != rgd->rd_rgl->rl_magic) { 1213 rgd->rd_rgl->rl_unlinked = cpu_to_be32(count_unlinked(rgd)); 1214 gfs2_rgrp_ondisk2lvb(rgd->rd_rgl, 1215 rgd->rd_bits[0].bi_bh->b_data); 1216 } 1217 else if (sdp->sd_args.ar_rgrplvb) { 1218 if (!gfs2_rgrp_lvb_valid(rgd)){ 1219 gfs2_consist_rgrpd(rgd); 1220 error = -EIO; 1221 goto fail; 1222 } 1223 if (rgd->rd_rgl->rl_unlinked == 0) 1224 rgd->rd_flags &= ~GFS2_RDF_CHECK; 1225 } 1226 return 0; 1227 1228 fail: 1229 while (x--) { 1230 bi = rgd->rd_bits + x; 1231 brelse(bi->bi_bh); 1232 bi->bi_bh = NULL; 1233 gfs2_assert_warn(sdp, !bi->bi_clone); 1234 } 1235 1236 return error; 1237 } 1238 1239 static int update_rgrp_lvb(struct gfs2_rgrpd *rgd) 1240 { 1241 u32 rl_flags; 1242 1243 if (rgd->rd_flags & GFS2_RDF_UPTODATE) 1244 return 0; 1245 1246 if (cpu_to_be32(GFS2_MAGIC) != rgd->rd_rgl->rl_magic) 1247 return gfs2_rgrp_bh_get(rgd); 1248 1249 rl_flags = be32_to_cpu(rgd->rd_rgl->rl_flags); 1250 rl_flags &= ~GFS2_RDF_MASK; 1251 rgd->rd_flags &= GFS2_RDF_MASK; 1252 rgd->rd_flags |= (rl_flags | GFS2_RDF_CHECK); 1253 if (rgd->rd_rgl->rl_unlinked == 0) 1254 rgd->rd_flags &= ~GFS2_RDF_CHECK; 1255 rgd->rd_free = be32_to_cpu(rgd->rd_rgl->rl_free); 1256 rgd->rd_free_clone = rgd->rd_free; 1257 rgd->rd_dinodes = be32_to_cpu(rgd->rd_rgl->rl_dinodes); 1258 rgd->rd_igeneration = be64_to_cpu(rgd->rd_rgl->rl_igeneration); 1259 return 0; 1260 } 1261 1262 int gfs2_rgrp_go_lock(struct gfs2_holder *gh) 1263 { 1264 struct gfs2_rgrpd *rgd = gh->gh_gl->gl_object; 1265 struct gfs2_sbd *sdp = rgd->rd_sbd; 1266 1267 if (gh->gh_flags & GL_SKIP && sdp->sd_args.ar_rgrplvb) 1268 return 0; 1269 return gfs2_rgrp_bh_get(rgd); 1270 } 1271 1272 /** 1273 * gfs2_rgrp_brelse - Release RG bitmaps read in with gfs2_rgrp_bh_get() 1274 * @rgd: The resource group 1275 * 1276 */ 1277 1278 void gfs2_rgrp_brelse(struct gfs2_rgrpd *rgd) 1279 { 1280 int x, length = rgd->rd_length; 1281 1282 for (x = 0; x < length; x++) { 1283 struct gfs2_bitmap *bi = rgd->rd_bits + x; 1284 if (bi->bi_bh) { 1285 brelse(bi->bi_bh); 1286 bi->bi_bh = NULL; 1287 } 1288 } 1289 1290 } 1291 1292 /** 1293 * gfs2_rgrp_go_unlock - Unlock a rgrp glock 1294 * @gh: The glock holder for the resource group 1295 * 1296 */ 1297 1298 void gfs2_rgrp_go_unlock(struct gfs2_holder *gh) 1299 { 1300 struct gfs2_rgrpd *rgd = gh->gh_gl->gl_object; 1301 int demote_requested = test_bit(GLF_DEMOTE, &gh->gh_gl->gl_flags) | 1302 test_bit(GLF_PENDING_DEMOTE, &gh->gh_gl->gl_flags); 1303 1304 if (rgd && demote_requested) 1305 gfs2_rgrp_brelse(rgd); 1306 } 1307 1308 int gfs2_rgrp_send_discards(struct gfs2_sbd *sdp, u64 offset, 1309 struct buffer_head *bh, 1310 const struct gfs2_bitmap *bi, unsigned minlen, u64 *ptrimmed) 1311 { 1312 struct super_block *sb = sdp->sd_vfs; 1313 u64 blk; 1314 sector_t start = 0; 1315 sector_t nr_blks = 0; 1316 int rv; 1317 unsigned int x; 1318 u32 trimmed = 0; 1319 u8 diff; 1320 1321 for (x = 0; x < bi->bi_bytes; x++) { 1322 const u8 *clone = bi->bi_clone ? bi->bi_clone : bi->bi_bh->b_data; 1323 clone += bi->bi_offset; 1324 clone += x; 1325 if (bh) { 1326 const u8 *orig = bh->b_data + bi->bi_offset + x; 1327 diff = ~(*orig | (*orig >> 1)) & (*clone | (*clone >> 1)); 1328 } else { 1329 diff = ~(*clone | (*clone >> 1)); 1330 } 1331 diff &= 0x55; 1332 if (diff == 0) 1333 continue; 1334 blk = offset + ((bi->bi_start + x) * GFS2_NBBY); 1335 while(diff) { 1336 if (diff & 1) { 1337 if (nr_blks == 0) 1338 goto start_new_extent; 1339 if ((start + nr_blks) != blk) { 1340 if (nr_blks >= minlen) { 1341 rv = sb_issue_discard(sb, 1342 start, nr_blks, 1343 GFP_NOFS, 0); 1344 if (rv) 1345 goto fail; 1346 trimmed += nr_blks; 1347 } 1348 nr_blks = 0; 1349 start_new_extent: 1350 start = blk; 1351 } 1352 nr_blks++; 1353 } 1354 diff >>= 2; 1355 blk++; 1356 } 1357 } 1358 if (nr_blks >= minlen) { 1359 rv = sb_issue_discard(sb, start, nr_blks, GFP_NOFS, 0); 1360 if (rv) 1361 goto fail; 1362 trimmed += nr_blks; 1363 } 1364 if (ptrimmed) 1365 *ptrimmed = trimmed; 1366 return 0; 1367 1368 fail: 1369 if (sdp->sd_args.ar_discard) 1370 fs_warn(sdp, "error %d on discard request, turning discards off for this filesystem\n", rv); 1371 sdp->sd_args.ar_discard = 0; 1372 return -EIO; 1373 } 1374 1375 /** 1376 * gfs2_fitrim - Generate discard requests for unused bits of the filesystem 1377 * @filp: Any file on the filesystem 1378 * @argp: Pointer to the arguments (also used to pass result) 1379 * 1380 * Returns: 0 on success, otherwise error code 1381 */ 1382 1383 int gfs2_fitrim(struct file *filp, void __user *argp) 1384 { 1385 struct inode *inode = file_inode(filp); 1386 struct gfs2_sbd *sdp = GFS2_SB(inode); 1387 struct request_queue *q = bdev_get_queue(sdp->sd_vfs->s_bdev); 1388 struct buffer_head *bh; 1389 struct gfs2_rgrpd *rgd; 1390 struct gfs2_rgrpd *rgd_end; 1391 struct gfs2_holder gh; 1392 struct fstrim_range r; 1393 int ret = 0; 1394 u64 amt; 1395 u64 trimmed = 0; 1396 u64 start, end, minlen; 1397 unsigned int x; 1398 unsigned bs_shift = sdp->sd_sb.sb_bsize_shift; 1399 1400 if (!capable(CAP_SYS_ADMIN)) 1401 return -EPERM; 1402 1403 if (!blk_queue_discard(q)) 1404 return -EOPNOTSUPP; 1405 1406 if (copy_from_user(&r, argp, sizeof(r))) 1407 return -EFAULT; 1408 1409 ret = gfs2_rindex_update(sdp); 1410 if (ret) 1411 return ret; 1412 1413 start = r.start >> bs_shift; 1414 end = start + (r.len >> bs_shift); 1415 minlen = max_t(u64, r.minlen, 1416 q->limits.discard_granularity) >> bs_shift; 1417 1418 if (end <= start || minlen > sdp->sd_max_rg_data) 1419 return -EINVAL; 1420 1421 rgd = gfs2_blk2rgrpd(sdp, start, 0); 1422 rgd_end = gfs2_blk2rgrpd(sdp, end, 0); 1423 1424 if ((gfs2_rgrpd_get_first(sdp) == gfs2_rgrpd_get_next(rgd_end)) 1425 && (start > rgd_end->rd_data0 + rgd_end->rd_data)) 1426 return -EINVAL; /* start is beyond the end of the fs */ 1427 1428 while (1) { 1429 1430 ret = gfs2_glock_nq_init(rgd->rd_gl, LM_ST_EXCLUSIVE, 0, &gh); 1431 if (ret) 1432 goto out; 1433 1434 if (!(rgd->rd_flags & GFS2_RGF_TRIMMED)) { 1435 /* Trim each bitmap in the rgrp */ 1436 for (x = 0; x < rgd->rd_length; x++) { 1437 struct gfs2_bitmap *bi = rgd->rd_bits + x; 1438 ret = gfs2_rgrp_send_discards(sdp, 1439 rgd->rd_data0, NULL, bi, minlen, 1440 &amt); 1441 if (ret) { 1442 gfs2_glock_dq_uninit(&gh); 1443 goto out; 1444 } 1445 trimmed += amt; 1446 } 1447 1448 /* Mark rgrp as having been trimmed */ 1449 ret = gfs2_trans_begin(sdp, RES_RG_HDR, 0); 1450 if (ret == 0) { 1451 bh = rgd->rd_bits[0].bi_bh; 1452 rgd->rd_flags |= GFS2_RGF_TRIMMED; 1453 gfs2_trans_add_meta(rgd->rd_gl, bh); 1454 gfs2_rgrp_out(rgd, bh->b_data); 1455 gfs2_trans_end(sdp); 1456 } 1457 } 1458 gfs2_glock_dq_uninit(&gh); 1459 1460 if (rgd == rgd_end) 1461 break; 1462 1463 rgd = gfs2_rgrpd_get_next(rgd); 1464 } 1465 1466 out: 1467 r.len = trimmed << bs_shift; 1468 if (copy_to_user(argp, &r, sizeof(r))) 1469 return -EFAULT; 1470 1471 return ret; 1472 } 1473 1474 /** 1475 * rs_insert - insert a new multi-block reservation into the rgrp's rb_tree 1476 * @ip: the inode structure 1477 * 1478 */ 1479 static void rs_insert(struct gfs2_inode *ip) 1480 { 1481 struct rb_node **newn, *parent = NULL; 1482 int rc; 1483 struct gfs2_blkreserv *rs = &ip->i_res; 1484 struct gfs2_rgrpd *rgd = rs->rs_rbm.rgd; 1485 u64 fsblock = gfs2_rbm_to_block(&rs->rs_rbm); 1486 1487 BUG_ON(gfs2_rs_active(rs)); 1488 1489 spin_lock(&rgd->rd_rsspin); 1490 newn = &rgd->rd_rstree.rb_node; 1491 while (*newn) { 1492 struct gfs2_blkreserv *cur = 1493 rb_entry(*newn, struct gfs2_blkreserv, rs_node); 1494 1495 parent = *newn; 1496 rc = rs_cmp(fsblock, rs->rs_free, cur); 1497 if (rc > 0) 1498 newn = &((*newn)->rb_right); 1499 else if (rc < 0) 1500 newn = &((*newn)->rb_left); 1501 else { 1502 spin_unlock(&rgd->rd_rsspin); 1503 WARN_ON(1); 1504 return; 1505 } 1506 } 1507 1508 rb_link_node(&rs->rs_node, parent, newn); 1509 rb_insert_color(&rs->rs_node, &rgd->rd_rstree); 1510 1511 /* Do our rgrp accounting for the reservation */ 1512 rgd->rd_reserved += rs->rs_free; /* blocks reserved */ 1513 spin_unlock(&rgd->rd_rsspin); 1514 trace_gfs2_rs(rs, TRACE_RS_INSERT); 1515 } 1516 1517 /** 1518 * rgd_free - return the number of free blocks we can allocate. 1519 * @rgd: the resource group 1520 * 1521 * This function returns the number of free blocks for an rgrp. 1522 * That's the clone-free blocks (blocks that are free, not including those 1523 * still being used for unlinked files that haven't been deleted.) 1524 * 1525 * It also subtracts any blocks reserved by someone else, but does not 1526 * include free blocks that are still part of our current reservation, 1527 * because obviously we can (and will) allocate them. 1528 */ 1529 static inline u32 rgd_free(struct gfs2_rgrpd *rgd, struct gfs2_blkreserv *rs) 1530 { 1531 u32 tot_reserved, tot_free; 1532 1533 if (WARN_ON_ONCE(rgd->rd_reserved < rs->rs_free)) 1534 return 0; 1535 tot_reserved = rgd->rd_reserved - rs->rs_free; 1536 1537 if (rgd->rd_free_clone < tot_reserved) 1538 tot_reserved = 0; 1539 1540 tot_free = rgd->rd_free_clone - tot_reserved; 1541 1542 return tot_free; 1543 } 1544 1545 /** 1546 * rg_mblk_search - find a group of multiple free blocks to form a reservation 1547 * @rgd: the resource group descriptor 1548 * @ip: pointer to the inode for which we're reserving blocks 1549 * @ap: the allocation parameters 1550 * 1551 */ 1552 1553 static void rg_mblk_search(struct gfs2_rgrpd *rgd, struct gfs2_inode *ip, 1554 const struct gfs2_alloc_parms *ap) 1555 { 1556 struct gfs2_rbm rbm = { .rgd = rgd, }; 1557 u64 goal; 1558 struct gfs2_blkreserv *rs = &ip->i_res; 1559 u32 extlen; 1560 u32 free_blocks = rgd_free(rgd, rs); 1561 int ret; 1562 struct inode *inode = &ip->i_inode; 1563 1564 if (S_ISDIR(inode->i_mode)) 1565 extlen = 1; 1566 else { 1567 extlen = max_t(u32, atomic_read(&ip->i_sizehint), ap->target); 1568 extlen = clamp(extlen, (u32)RGRP_RSRV_MINBLKS, free_blocks); 1569 } 1570 if ((rgd->rd_free_clone < rgd->rd_reserved) || (free_blocks < extlen)) 1571 return; 1572 1573 /* Find bitmap block that contains bits for goal block */ 1574 if (rgrp_contains_block(rgd, ip->i_goal)) 1575 goal = ip->i_goal; 1576 else 1577 goal = rgd->rd_last_alloc + rgd->rd_data0; 1578 1579 if (WARN_ON(gfs2_rbm_from_block(&rbm, goal))) 1580 return; 1581 1582 ret = gfs2_rbm_find(&rbm, GFS2_BLKST_FREE, &extlen, ip, true); 1583 if (ret == 0) { 1584 rs->rs_rbm = rbm; 1585 rs->rs_free = extlen; 1586 rs_insert(ip); 1587 } else { 1588 if (goal == rgd->rd_last_alloc + rgd->rd_data0) 1589 rgd->rd_last_alloc = 0; 1590 } 1591 } 1592 1593 /** 1594 * gfs2_next_unreserved_block - Return next block that is not reserved 1595 * @rgd: The resource group 1596 * @block: The starting block 1597 * @length: The required length 1598 * @ip: Ignore any reservations for this inode 1599 * 1600 * If the block does not appear in any reservation, then return the 1601 * block number unchanged. If it does appear in the reservation, then 1602 * keep looking through the tree of reservations in order to find the 1603 * first block number which is not reserved. 1604 */ 1605 1606 static u64 gfs2_next_unreserved_block(struct gfs2_rgrpd *rgd, u64 block, 1607 u32 length, 1608 const struct gfs2_inode *ip) 1609 { 1610 struct gfs2_blkreserv *rs; 1611 struct rb_node *n; 1612 int rc; 1613 1614 spin_lock(&rgd->rd_rsspin); 1615 n = rgd->rd_rstree.rb_node; 1616 while (n) { 1617 rs = rb_entry(n, struct gfs2_blkreserv, rs_node); 1618 rc = rs_cmp(block, length, rs); 1619 if (rc < 0) 1620 n = n->rb_left; 1621 else if (rc > 0) 1622 n = n->rb_right; 1623 else 1624 break; 1625 } 1626 1627 if (n) { 1628 while ((rs_cmp(block, length, rs) == 0) && (&ip->i_res != rs)) { 1629 block = gfs2_rbm_to_block(&rs->rs_rbm) + rs->rs_free; 1630 n = n->rb_right; 1631 if (n == NULL) 1632 break; 1633 rs = rb_entry(n, struct gfs2_blkreserv, rs_node); 1634 } 1635 } 1636 1637 spin_unlock(&rgd->rd_rsspin); 1638 return block; 1639 } 1640 1641 /** 1642 * gfs2_reservation_check_and_update - Check for reservations during block alloc 1643 * @rbm: The current position in the resource group 1644 * @ip: The inode for which we are searching for blocks 1645 * @minext: The minimum extent length 1646 * @maxext: A pointer to the maximum extent structure 1647 * 1648 * This checks the current position in the rgrp to see whether there is 1649 * a reservation covering this block. If not then this function is a 1650 * no-op. If there is, then the position is moved to the end of the 1651 * contiguous reservation(s) so that we are pointing at the first 1652 * non-reserved block. 1653 * 1654 * Returns: 0 if no reservation, 1 if @rbm has changed, otherwise an error 1655 */ 1656 1657 static int gfs2_reservation_check_and_update(struct gfs2_rbm *rbm, 1658 const struct gfs2_inode *ip, 1659 u32 minext, 1660 struct gfs2_extent *maxext) 1661 { 1662 u64 block = gfs2_rbm_to_block(rbm); 1663 u32 extlen = 1; 1664 u64 nblock; 1665 int ret; 1666 1667 /* 1668 * If we have a minimum extent length, then skip over any extent 1669 * which is less than the min extent length in size. 1670 */ 1671 if (minext) { 1672 extlen = gfs2_free_extlen(rbm, minext); 1673 if (extlen <= maxext->len) 1674 goto fail; 1675 } 1676 1677 /* 1678 * Check the extent which has been found against the reservations 1679 * and skip if parts of it are already reserved 1680 */ 1681 nblock = gfs2_next_unreserved_block(rbm->rgd, block, extlen, ip); 1682 if (nblock == block) { 1683 if (!minext || extlen >= minext) 1684 return 0; 1685 1686 if (extlen > maxext->len) { 1687 maxext->len = extlen; 1688 maxext->rbm = *rbm; 1689 } 1690 fail: 1691 nblock = block + extlen; 1692 } 1693 ret = gfs2_rbm_from_block(rbm, nblock); 1694 if (ret < 0) 1695 return ret; 1696 return 1; 1697 } 1698 1699 /** 1700 * gfs2_rbm_find - Look for blocks of a particular state 1701 * @rbm: Value/result starting position and final position 1702 * @state: The state which we want to find 1703 * @minext: Pointer to the requested extent length (NULL for a single block) 1704 * This is updated to be the actual reservation size. 1705 * @ip: If set, check for reservations 1706 * @nowrap: Stop looking at the end of the rgrp, rather than wrapping 1707 * around until we've reached the starting point. 1708 * 1709 * Side effects: 1710 * - If looking for free blocks, we set GBF_FULL on each bitmap which 1711 * has no free blocks in it. 1712 * - If looking for free blocks, we set rd_extfail_pt on each rgrp which 1713 * has come up short on a free block search. 1714 * 1715 * Returns: 0 on success, -ENOSPC if there is no block of the requested state 1716 */ 1717 1718 static int gfs2_rbm_find(struct gfs2_rbm *rbm, u8 state, u32 *minext, 1719 const struct gfs2_inode *ip, bool nowrap) 1720 { 1721 bool scan_from_start = rbm->bii == 0 && rbm->offset == 0; 1722 struct buffer_head *bh; 1723 int last_bii; 1724 u32 offset; 1725 u8 *buffer; 1726 bool wrapped = false; 1727 int ret; 1728 struct gfs2_bitmap *bi; 1729 struct gfs2_extent maxext = { .rbm.rgd = rbm->rgd, }; 1730 1731 /* 1732 * Determine the last bitmap to search. If we're not starting at the 1733 * beginning of a bitmap, we need to search that bitmap twice to scan 1734 * the entire resource group. 1735 */ 1736 last_bii = rbm->bii - (rbm->offset == 0); 1737 1738 while(1) { 1739 bi = rbm_bi(rbm); 1740 if ((ip == NULL || !gfs2_rs_active(&ip->i_res)) && 1741 test_bit(GBF_FULL, &bi->bi_flags) && 1742 (state == GFS2_BLKST_FREE)) 1743 goto next_bitmap; 1744 1745 bh = bi->bi_bh; 1746 buffer = bh->b_data + bi->bi_offset; 1747 WARN_ON(!buffer_uptodate(bh)); 1748 if (state != GFS2_BLKST_UNLINKED && bi->bi_clone) 1749 buffer = bi->bi_clone + bi->bi_offset; 1750 offset = gfs2_bitfit(buffer, bi->bi_bytes, rbm->offset, state); 1751 if (offset == BFITNOENT) { 1752 if (state == GFS2_BLKST_FREE && rbm->offset == 0) 1753 set_bit(GBF_FULL, &bi->bi_flags); 1754 goto next_bitmap; 1755 } 1756 rbm->offset = offset; 1757 if (ip == NULL) 1758 return 0; 1759 1760 ret = gfs2_reservation_check_and_update(rbm, ip, 1761 minext ? *minext : 0, 1762 &maxext); 1763 if (ret == 0) 1764 return 0; 1765 if (ret > 0) 1766 goto next_iter; 1767 if (ret == -E2BIG) { 1768 rbm->bii = 0; 1769 rbm->offset = 0; 1770 goto res_covered_end_of_rgrp; 1771 } 1772 return ret; 1773 1774 next_bitmap: /* Find next bitmap in the rgrp */ 1775 rbm->offset = 0; 1776 rbm->bii++; 1777 if (rbm->bii == rbm->rgd->rd_length) 1778 rbm->bii = 0; 1779 res_covered_end_of_rgrp: 1780 if (rbm->bii == 0) { 1781 if (wrapped) 1782 break; 1783 wrapped = true; 1784 if (nowrap) 1785 break; 1786 } 1787 next_iter: 1788 /* Have we scanned the entire resource group? */ 1789 if (wrapped && rbm->bii > last_bii) 1790 break; 1791 } 1792 1793 if (minext == NULL || state != GFS2_BLKST_FREE) 1794 return -ENOSPC; 1795 1796 /* If the extent was too small, and it's smaller than the smallest 1797 to have failed before, remember for future reference that it's 1798 useless to search this rgrp again for this amount or more. */ 1799 if (wrapped && (scan_from_start || rbm->bii > last_bii) && 1800 *minext < rbm->rgd->rd_extfail_pt) 1801 rbm->rgd->rd_extfail_pt = *minext; 1802 1803 /* If the maximum extent we found is big enough to fulfill the 1804 minimum requirements, use it anyway. */ 1805 if (maxext.len) { 1806 *rbm = maxext.rbm; 1807 *minext = maxext.len; 1808 return 0; 1809 } 1810 1811 return -ENOSPC; 1812 } 1813 1814 /** 1815 * try_rgrp_unlink - Look for any unlinked, allocated, but unused inodes 1816 * @rgd: The rgrp 1817 * @last_unlinked: block address of the last dinode we unlinked 1818 * @skip: block address we should explicitly not unlink 1819 * 1820 * Returns: 0 if no error 1821 * The inode, if one has been found, in inode. 1822 */ 1823 1824 static void try_rgrp_unlink(struct gfs2_rgrpd *rgd, u64 *last_unlinked, u64 skip) 1825 { 1826 u64 block; 1827 struct gfs2_sbd *sdp = rgd->rd_sbd; 1828 struct gfs2_glock *gl; 1829 struct gfs2_inode *ip; 1830 int error; 1831 int found = 0; 1832 struct gfs2_rbm rbm = { .rgd = rgd, .bii = 0, .offset = 0 }; 1833 1834 while (1) { 1835 down_write(&sdp->sd_log_flush_lock); 1836 error = gfs2_rbm_find(&rbm, GFS2_BLKST_UNLINKED, NULL, NULL, 1837 true); 1838 up_write(&sdp->sd_log_flush_lock); 1839 if (error == -ENOSPC) 1840 break; 1841 if (WARN_ON_ONCE(error)) 1842 break; 1843 1844 block = gfs2_rbm_to_block(&rbm); 1845 if (gfs2_rbm_from_block(&rbm, block + 1)) 1846 break; 1847 if (*last_unlinked != NO_BLOCK && block <= *last_unlinked) 1848 continue; 1849 if (block == skip) 1850 continue; 1851 *last_unlinked = block; 1852 1853 error = gfs2_glock_get(sdp, block, &gfs2_iopen_glops, CREATE, &gl); 1854 if (error) 1855 continue; 1856 1857 /* If the inode is already in cache, we can ignore it here 1858 * because the existing inode disposal code will deal with 1859 * it when all refs have gone away. Accessing gl_object like 1860 * this is not safe in general. Here it is ok because we do 1861 * not dereference the pointer, and we only need an approx 1862 * answer to whether it is NULL or not. 1863 */ 1864 ip = gl->gl_object; 1865 1866 if (ip || queue_work(gfs2_delete_workqueue, &gl->gl_delete) == 0) 1867 gfs2_glock_put(gl); 1868 else 1869 found++; 1870 1871 /* Limit reclaim to sensible number of tasks */ 1872 if (found > NR_CPUS) 1873 return; 1874 } 1875 1876 rgd->rd_flags &= ~GFS2_RDF_CHECK; 1877 return; 1878 } 1879 1880 /** 1881 * gfs2_rgrp_congested - Use stats to figure out whether an rgrp is congested 1882 * @rgd: The rgrp in question 1883 * @loops: An indication of how picky we can be (0=very, 1=less so) 1884 * 1885 * This function uses the recently added glock statistics in order to 1886 * figure out whether a parciular resource group is suffering from 1887 * contention from multiple nodes. This is done purely on the basis 1888 * of timings, since this is the only data we have to work with and 1889 * our aim here is to reject a resource group which is highly contended 1890 * but (very important) not to do this too often in order to ensure that 1891 * we do not land up introducing fragmentation by changing resource 1892 * groups when not actually required. 1893 * 1894 * The calculation is fairly simple, we want to know whether the SRTTB 1895 * (i.e. smoothed round trip time for blocking operations) to acquire 1896 * the lock for this rgrp's glock is significantly greater than the 1897 * time taken for resource groups on average. We introduce a margin in 1898 * the form of the variable @var which is computed as the sum of the two 1899 * respective variences, and multiplied by a factor depending on @loops 1900 * and whether we have a lot of data to base the decision on. This is 1901 * then tested against the square difference of the means in order to 1902 * decide whether the result is statistically significant or not. 1903 * 1904 * Returns: A boolean verdict on the congestion status 1905 */ 1906 1907 static bool gfs2_rgrp_congested(const struct gfs2_rgrpd *rgd, int loops) 1908 { 1909 const struct gfs2_glock *gl = rgd->rd_gl; 1910 const struct gfs2_sbd *sdp = gl->gl_name.ln_sbd; 1911 struct gfs2_lkstats *st; 1912 u64 r_dcount, l_dcount; 1913 u64 l_srttb, a_srttb = 0; 1914 s64 srttb_diff; 1915 u64 sqr_diff; 1916 u64 var; 1917 int cpu, nonzero = 0; 1918 1919 preempt_disable(); 1920 for_each_present_cpu(cpu) { 1921 st = &per_cpu_ptr(sdp->sd_lkstats, cpu)->lkstats[LM_TYPE_RGRP]; 1922 if (st->stats[GFS2_LKS_SRTTB]) { 1923 a_srttb += st->stats[GFS2_LKS_SRTTB]; 1924 nonzero++; 1925 } 1926 } 1927 st = &this_cpu_ptr(sdp->sd_lkstats)->lkstats[LM_TYPE_RGRP]; 1928 if (nonzero) 1929 do_div(a_srttb, nonzero); 1930 r_dcount = st->stats[GFS2_LKS_DCOUNT]; 1931 var = st->stats[GFS2_LKS_SRTTVARB] + 1932 gl->gl_stats.stats[GFS2_LKS_SRTTVARB]; 1933 preempt_enable(); 1934 1935 l_srttb = gl->gl_stats.stats[GFS2_LKS_SRTTB]; 1936 l_dcount = gl->gl_stats.stats[GFS2_LKS_DCOUNT]; 1937 1938 if ((l_dcount < 1) || (r_dcount < 1) || (a_srttb == 0)) 1939 return false; 1940 1941 srttb_diff = a_srttb - l_srttb; 1942 sqr_diff = srttb_diff * srttb_diff; 1943 1944 var *= 2; 1945 if (l_dcount < 8 || r_dcount < 8) 1946 var *= 2; 1947 if (loops == 1) 1948 var *= 2; 1949 1950 return ((srttb_diff < 0) && (sqr_diff > var)); 1951 } 1952 1953 /** 1954 * gfs2_rgrp_used_recently 1955 * @rs: The block reservation with the rgrp to test 1956 * @msecs: The time limit in milliseconds 1957 * 1958 * Returns: True if the rgrp glock has been used within the time limit 1959 */ 1960 static bool gfs2_rgrp_used_recently(const struct gfs2_blkreserv *rs, 1961 u64 msecs) 1962 { 1963 u64 tdiff; 1964 1965 tdiff = ktime_to_ns(ktime_sub(ktime_get_real(), 1966 rs->rs_rbm.rgd->rd_gl->gl_dstamp)); 1967 1968 return tdiff > (msecs * 1000 * 1000); 1969 } 1970 1971 static u32 gfs2_orlov_skip(const struct gfs2_inode *ip) 1972 { 1973 const struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 1974 u32 skip; 1975 1976 get_random_bytes(&skip, sizeof(skip)); 1977 return skip % sdp->sd_rgrps; 1978 } 1979 1980 static bool gfs2_select_rgrp(struct gfs2_rgrpd **pos, const struct gfs2_rgrpd *begin) 1981 { 1982 struct gfs2_rgrpd *rgd = *pos; 1983 struct gfs2_sbd *sdp = rgd->rd_sbd; 1984 1985 rgd = gfs2_rgrpd_get_next(rgd); 1986 if (rgd == NULL) 1987 rgd = gfs2_rgrpd_get_first(sdp); 1988 *pos = rgd; 1989 if (rgd != begin) /* If we didn't wrap */ 1990 return true; 1991 return false; 1992 } 1993 1994 /** 1995 * fast_to_acquire - determine if a resource group will be fast to acquire 1996 * 1997 * If this is one of our preferred rgrps, it should be quicker to acquire, 1998 * because we tried to set ourselves up as dlm lock master. 1999 */ 2000 static inline int fast_to_acquire(struct gfs2_rgrpd *rgd) 2001 { 2002 struct gfs2_glock *gl = rgd->rd_gl; 2003 2004 if (gl->gl_state != LM_ST_UNLOCKED && list_empty(&gl->gl_holders) && 2005 !test_bit(GLF_DEMOTE_IN_PROGRESS, &gl->gl_flags) && 2006 !test_bit(GLF_DEMOTE, &gl->gl_flags)) 2007 return 1; 2008 if (rgd->rd_flags & GFS2_RDF_PREFERRED) 2009 return 1; 2010 return 0; 2011 } 2012 2013 /** 2014 * gfs2_inplace_reserve - Reserve space in the filesystem 2015 * @ip: the inode to reserve space for 2016 * @ap: the allocation parameters 2017 * 2018 * We try our best to find an rgrp that has at least ap->target blocks 2019 * available. After a couple of passes (loops == 2), the prospects of finding 2020 * such an rgrp diminish. At this stage, we return the first rgrp that has 2021 * at least ap->min_target blocks available. Either way, we set ap->allowed to 2022 * the number of blocks available in the chosen rgrp. 2023 * 2024 * Returns: 0 on success, 2025 * -ENOMEM if a suitable rgrp can't be found 2026 * errno otherwise 2027 */ 2028 2029 int gfs2_inplace_reserve(struct gfs2_inode *ip, struct gfs2_alloc_parms *ap) 2030 { 2031 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 2032 struct gfs2_rgrpd *begin = NULL; 2033 struct gfs2_blkreserv *rs = &ip->i_res; 2034 int error = 0, rg_locked, flags = 0; 2035 u64 last_unlinked = NO_BLOCK; 2036 int loops = 0; 2037 u32 free_blocks, skip = 0; 2038 2039 if (sdp->sd_args.ar_rgrplvb) 2040 flags |= GL_SKIP; 2041 if (gfs2_assert_warn(sdp, ap->target)) 2042 return -EINVAL; 2043 if (gfs2_rs_active(rs)) { 2044 begin = rs->rs_rbm.rgd; 2045 } else if (rs->rs_rbm.rgd && 2046 rgrp_contains_block(rs->rs_rbm.rgd, ip->i_goal)) { 2047 begin = rs->rs_rbm.rgd; 2048 } else { 2049 check_and_update_goal(ip); 2050 rs->rs_rbm.rgd = begin = gfs2_blk2rgrpd(sdp, ip->i_goal, 1); 2051 } 2052 if (S_ISDIR(ip->i_inode.i_mode) && (ap->aflags & GFS2_AF_ORLOV)) 2053 skip = gfs2_orlov_skip(ip); 2054 if (rs->rs_rbm.rgd == NULL) 2055 return -EBADSLT; 2056 2057 while (loops < 3) { 2058 rg_locked = 1; 2059 2060 if (!gfs2_glock_is_locked_by_me(rs->rs_rbm.rgd->rd_gl)) { 2061 rg_locked = 0; 2062 if (skip && skip--) 2063 goto next_rgrp; 2064 if (!gfs2_rs_active(rs)) { 2065 if (loops == 0 && 2066 !fast_to_acquire(rs->rs_rbm.rgd)) 2067 goto next_rgrp; 2068 if ((loops < 2) && 2069 gfs2_rgrp_used_recently(rs, 1000) && 2070 gfs2_rgrp_congested(rs->rs_rbm.rgd, loops)) 2071 goto next_rgrp; 2072 } 2073 error = gfs2_glock_nq_init(rs->rs_rbm.rgd->rd_gl, 2074 LM_ST_EXCLUSIVE, flags, 2075 &ip->i_rgd_gh); 2076 if (unlikely(error)) 2077 return error; 2078 if (!gfs2_rs_active(rs) && (loops < 2) && 2079 gfs2_rgrp_congested(rs->rs_rbm.rgd, loops)) 2080 goto skip_rgrp; 2081 if (sdp->sd_args.ar_rgrplvb) { 2082 error = update_rgrp_lvb(rs->rs_rbm.rgd); 2083 if (unlikely(error)) { 2084 gfs2_glock_dq_uninit(&ip->i_rgd_gh); 2085 return error; 2086 } 2087 } 2088 } 2089 2090 /* Skip unusable resource groups */ 2091 if ((rs->rs_rbm.rgd->rd_flags & (GFS2_RGF_NOALLOC | 2092 GFS2_RDF_ERROR)) || 2093 (loops == 0 && ap->target > rs->rs_rbm.rgd->rd_extfail_pt)) 2094 goto skip_rgrp; 2095 2096 if (sdp->sd_args.ar_rgrplvb) 2097 gfs2_rgrp_bh_get(rs->rs_rbm.rgd); 2098 2099 /* Get a reservation if we don't already have one */ 2100 if (!gfs2_rs_active(rs)) 2101 rg_mblk_search(rs->rs_rbm.rgd, ip, ap); 2102 2103 /* Skip rgrps when we can't get a reservation on first pass */ 2104 if (!gfs2_rs_active(rs) && (loops < 1)) 2105 goto check_rgrp; 2106 2107 /* If rgrp has enough free space, use it */ 2108 free_blocks = rgd_free(rs->rs_rbm.rgd, rs); 2109 if (free_blocks >= ap->target || 2110 (loops == 2 && ap->min_target && 2111 free_blocks >= ap->min_target)) { 2112 ap->allowed = free_blocks; 2113 return 0; 2114 } 2115 check_rgrp: 2116 /* Check for unlinked inodes which can be reclaimed */ 2117 if (rs->rs_rbm.rgd->rd_flags & GFS2_RDF_CHECK) 2118 try_rgrp_unlink(rs->rs_rbm.rgd, &last_unlinked, 2119 ip->i_no_addr); 2120 skip_rgrp: 2121 /* Drop reservation, if we couldn't use reserved rgrp */ 2122 if (gfs2_rs_active(rs)) 2123 gfs2_rs_deltree(rs); 2124 2125 /* Unlock rgrp if required */ 2126 if (!rg_locked) 2127 gfs2_glock_dq_uninit(&ip->i_rgd_gh); 2128 next_rgrp: 2129 /* Find the next rgrp, and continue looking */ 2130 if (gfs2_select_rgrp(&rs->rs_rbm.rgd, begin)) 2131 continue; 2132 if (skip) 2133 continue; 2134 2135 /* If we've scanned all the rgrps, but found no free blocks 2136 * then this checks for some less likely conditions before 2137 * trying again. 2138 */ 2139 loops++; 2140 /* Check that fs hasn't grown if writing to rindex */ 2141 if (ip == GFS2_I(sdp->sd_rindex) && !sdp->sd_rindex_uptodate) { 2142 error = gfs2_ri_update(ip); 2143 if (error) 2144 return error; 2145 } 2146 /* Flushing the log may release space */ 2147 if (loops == 2) 2148 gfs2_log_flush(sdp, NULL, GFS2_LOG_HEAD_FLUSH_NORMAL | 2149 GFS2_LFC_INPLACE_RESERVE); 2150 } 2151 2152 return -ENOSPC; 2153 } 2154 2155 /** 2156 * gfs2_inplace_release - release an inplace reservation 2157 * @ip: the inode the reservation was taken out on 2158 * 2159 * Release a reservation made by gfs2_inplace_reserve(). 2160 */ 2161 2162 void gfs2_inplace_release(struct gfs2_inode *ip) 2163 { 2164 if (gfs2_holder_initialized(&ip->i_rgd_gh)) 2165 gfs2_glock_dq_uninit(&ip->i_rgd_gh); 2166 } 2167 2168 /** 2169 * gfs2_alloc_extent - allocate an extent from a given bitmap 2170 * @rbm: the resource group information 2171 * @dinode: TRUE if the first block we allocate is for a dinode 2172 * @n: The extent length (value/result) 2173 * 2174 * Add the bitmap buffer to the transaction. 2175 * Set the found bits to @new_state to change block's allocation state. 2176 */ 2177 static void gfs2_alloc_extent(const struct gfs2_rbm *rbm, bool dinode, 2178 unsigned int *n) 2179 { 2180 struct gfs2_rbm pos = { .rgd = rbm->rgd, }; 2181 const unsigned int elen = *n; 2182 u64 block; 2183 int ret; 2184 2185 *n = 1; 2186 block = gfs2_rbm_to_block(rbm); 2187 gfs2_trans_add_meta(rbm->rgd->rd_gl, rbm_bi(rbm)->bi_bh); 2188 gfs2_setbit(rbm, true, dinode ? GFS2_BLKST_DINODE : GFS2_BLKST_USED); 2189 block++; 2190 while (*n < elen) { 2191 ret = gfs2_rbm_from_block(&pos, block); 2192 if (ret || gfs2_testbit(&pos, true) != GFS2_BLKST_FREE) 2193 break; 2194 gfs2_trans_add_meta(pos.rgd->rd_gl, rbm_bi(&pos)->bi_bh); 2195 gfs2_setbit(&pos, true, GFS2_BLKST_USED); 2196 (*n)++; 2197 block++; 2198 } 2199 } 2200 2201 /** 2202 * rgblk_free - Change alloc state of given block(s) 2203 * @sdp: the filesystem 2204 * @rgd: the resource group the blocks are in 2205 * @bstart: the start of a run of blocks to free 2206 * @blen: the length of the block run (all must lie within ONE RG!) 2207 * @new_state: GFS2_BLKST_XXX the after-allocation block state 2208 */ 2209 2210 static void rgblk_free(struct gfs2_sbd *sdp, struct gfs2_rgrpd *rgd, 2211 u64 bstart, u32 blen, unsigned char new_state) 2212 { 2213 struct gfs2_rbm rbm; 2214 struct gfs2_bitmap *bi, *bi_prev = NULL; 2215 2216 rbm.rgd = rgd; 2217 if (WARN_ON_ONCE(gfs2_rbm_from_block(&rbm, bstart))) 2218 return; 2219 while (blen--) { 2220 bi = rbm_bi(&rbm); 2221 if (bi != bi_prev) { 2222 if (!bi->bi_clone) { 2223 bi->bi_clone = kmalloc(bi->bi_bh->b_size, 2224 GFP_NOFS | __GFP_NOFAIL); 2225 memcpy(bi->bi_clone + bi->bi_offset, 2226 bi->bi_bh->b_data + bi->bi_offset, 2227 bi->bi_bytes); 2228 } 2229 gfs2_trans_add_meta(rbm.rgd->rd_gl, bi->bi_bh); 2230 bi_prev = bi; 2231 } 2232 gfs2_setbit(&rbm, false, new_state); 2233 gfs2_rbm_incr(&rbm); 2234 } 2235 } 2236 2237 /** 2238 * gfs2_rgrp_dump - print out an rgrp 2239 * @seq: The iterator 2240 * @gl: The glock in question 2241 * @fs_id_buf: pointer to file system id (if requested) 2242 * 2243 */ 2244 2245 void gfs2_rgrp_dump(struct seq_file *seq, struct gfs2_glock *gl, 2246 const char *fs_id_buf) 2247 { 2248 struct gfs2_rgrpd *rgd = gl->gl_object; 2249 struct gfs2_blkreserv *trs; 2250 const struct rb_node *n; 2251 2252 if (rgd == NULL) 2253 return; 2254 gfs2_print_dbg(seq, "%s R: n:%llu f:%02x b:%u/%u i:%u r:%u e:%u\n", 2255 fs_id_buf, 2256 (unsigned long long)rgd->rd_addr, rgd->rd_flags, 2257 rgd->rd_free, rgd->rd_free_clone, rgd->rd_dinodes, 2258 rgd->rd_reserved, rgd->rd_extfail_pt); 2259 if (rgd->rd_sbd->sd_args.ar_rgrplvb) { 2260 struct gfs2_rgrp_lvb *rgl = rgd->rd_rgl; 2261 2262 gfs2_print_dbg(seq, "%s L: f:%02x b:%u i:%u\n", fs_id_buf, 2263 be32_to_cpu(rgl->rl_flags), 2264 be32_to_cpu(rgl->rl_free), 2265 be32_to_cpu(rgl->rl_dinodes)); 2266 } 2267 spin_lock(&rgd->rd_rsspin); 2268 for (n = rb_first(&rgd->rd_rstree); n; n = rb_next(&trs->rs_node)) { 2269 trs = rb_entry(n, struct gfs2_blkreserv, rs_node); 2270 dump_rs(seq, trs, fs_id_buf); 2271 } 2272 spin_unlock(&rgd->rd_rsspin); 2273 } 2274 2275 static void gfs2_rgrp_error(struct gfs2_rgrpd *rgd) 2276 { 2277 struct gfs2_sbd *sdp = rgd->rd_sbd; 2278 char fs_id_buf[sizeof(sdp->sd_fsname) + 7]; 2279 2280 fs_warn(sdp, "rgrp %llu has an error, marking it readonly until umount\n", 2281 (unsigned long long)rgd->rd_addr); 2282 fs_warn(sdp, "umount on all nodes and run fsck.gfs2 to fix the error\n"); 2283 sprintf(fs_id_buf, "fsid=%s: ", sdp->sd_fsname); 2284 gfs2_rgrp_dump(NULL, rgd->rd_gl, fs_id_buf); 2285 rgd->rd_flags |= GFS2_RDF_ERROR; 2286 } 2287 2288 /** 2289 * gfs2_adjust_reservation - Adjust (or remove) a reservation after allocation 2290 * @ip: The inode we have just allocated blocks for 2291 * @rbm: The start of the allocated blocks 2292 * @len: The extent length 2293 * 2294 * Adjusts a reservation after an allocation has taken place. If the 2295 * reservation does not match the allocation, or if it is now empty 2296 * then it is removed. 2297 */ 2298 2299 static void gfs2_adjust_reservation(struct gfs2_inode *ip, 2300 const struct gfs2_rbm *rbm, unsigned len) 2301 { 2302 struct gfs2_blkreserv *rs = &ip->i_res; 2303 struct gfs2_rgrpd *rgd = rbm->rgd; 2304 unsigned rlen; 2305 u64 block; 2306 int ret; 2307 2308 spin_lock(&rgd->rd_rsspin); 2309 if (gfs2_rs_active(rs)) { 2310 if (gfs2_rbm_eq(&rs->rs_rbm, rbm)) { 2311 block = gfs2_rbm_to_block(rbm); 2312 ret = gfs2_rbm_from_block(&rs->rs_rbm, block + len); 2313 rlen = min(rs->rs_free, len); 2314 rs->rs_free -= rlen; 2315 rgd->rd_reserved -= rlen; 2316 trace_gfs2_rs(rs, TRACE_RS_CLAIM); 2317 if (rs->rs_free && !ret) 2318 goto out; 2319 /* We used up our block reservation, so we should 2320 reserve more blocks next time. */ 2321 atomic_add(RGRP_RSRV_ADDBLKS, &ip->i_sizehint); 2322 } 2323 __rs_deltree(rs); 2324 } 2325 out: 2326 spin_unlock(&rgd->rd_rsspin); 2327 } 2328 2329 /** 2330 * gfs2_set_alloc_start - Set starting point for block allocation 2331 * @rbm: The rbm which will be set to the required location 2332 * @ip: The gfs2 inode 2333 * @dinode: Flag to say if allocation includes a new inode 2334 * 2335 * This sets the starting point from the reservation if one is active 2336 * otherwise it falls back to guessing a start point based on the 2337 * inode's goal block or the last allocation point in the rgrp. 2338 */ 2339 2340 static void gfs2_set_alloc_start(struct gfs2_rbm *rbm, 2341 const struct gfs2_inode *ip, bool dinode) 2342 { 2343 u64 goal; 2344 2345 if (gfs2_rs_active(&ip->i_res)) { 2346 *rbm = ip->i_res.rs_rbm; 2347 return; 2348 } 2349 2350 if (!dinode && rgrp_contains_block(rbm->rgd, ip->i_goal)) 2351 goal = ip->i_goal; 2352 else 2353 goal = rbm->rgd->rd_last_alloc + rbm->rgd->rd_data0; 2354 2355 if (WARN_ON_ONCE(gfs2_rbm_from_block(rbm, goal))) { 2356 rbm->bii = 0; 2357 rbm->offset = 0; 2358 } 2359 } 2360 2361 /** 2362 * gfs2_alloc_blocks - Allocate one or more blocks of data and/or a dinode 2363 * @ip: the inode to allocate the block for 2364 * @bn: Used to return the starting block number 2365 * @nblocks: requested number of blocks/extent length (value/result) 2366 * @dinode: 1 if we're allocating a dinode block, else 0 2367 * @generation: the generation number of the inode 2368 * 2369 * Returns: 0 or error 2370 */ 2371 2372 int gfs2_alloc_blocks(struct gfs2_inode *ip, u64 *bn, unsigned int *nblocks, 2373 bool dinode, u64 *generation) 2374 { 2375 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 2376 struct buffer_head *dibh; 2377 struct gfs2_rbm rbm = { .rgd = ip->i_res.rs_rbm.rgd, }; 2378 unsigned int ndata; 2379 u64 block; /* block, within the file system scope */ 2380 int error; 2381 2382 gfs2_set_alloc_start(&rbm, ip, dinode); 2383 error = gfs2_rbm_find(&rbm, GFS2_BLKST_FREE, NULL, ip, false); 2384 2385 if (error == -ENOSPC) { 2386 gfs2_set_alloc_start(&rbm, ip, dinode); 2387 error = gfs2_rbm_find(&rbm, GFS2_BLKST_FREE, NULL, NULL, false); 2388 } 2389 2390 /* Since all blocks are reserved in advance, this shouldn't happen */ 2391 if (error) { 2392 fs_warn(sdp, "inum=%llu error=%d, nblocks=%u, full=%d fail_pt=%d\n", 2393 (unsigned long long)ip->i_no_addr, error, *nblocks, 2394 test_bit(GBF_FULL, &rbm.rgd->rd_bits->bi_flags), 2395 rbm.rgd->rd_extfail_pt); 2396 goto rgrp_error; 2397 } 2398 2399 gfs2_alloc_extent(&rbm, dinode, nblocks); 2400 block = gfs2_rbm_to_block(&rbm); 2401 rbm.rgd->rd_last_alloc = block - rbm.rgd->rd_data0; 2402 if (gfs2_rs_active(&ip->i_res)) 2403 gfs2_adjust_reservation(ip, &rbm, *nblocks); 2404 ndata = *nblocks; 2405 if (dinode) 2406 ndata--; 2407 2408 if (!dinode) { 2409 ip->i_goal = block + ndata - 1; 2410 error = gfs2_meta_inode_buffer(ip, &dibh); 2411 if (error == 0) { 2412 struct gfs2_dinode *di = 2413 (struct gfs2_dinode *)dibh->b_data; 2414 gfs2_trans_add_meta(ip->i_gl, dibh); 2415 di->di_goal_meta = di->di_goal_data = 2416 cpu_to_be64(ip->i_goal); 2417 brelse(dibh); 2418 } 2419 } 2420 if (rbm.rgd->rd_free < *nblocks) { 2421 fs_warn(sdp, "nblocks=%u\n", *nblocks); 2422 goto rgrp_error; 2423 } 2424 2425 rbm.rgd->rd_free -= *nblocks; 2426 if (dinode) { 2427 rbm.rgd->rd_dinodes++; 2428 *generation = rbm.rgd->rd_igeneration++; 2429 if (*generation == 0) 2430 *generation = rbm.rgd->rd_igeneration++; 2431 } 2432 2433 gfs2_trans_add_meta(rbm.rgd->rd_gl, rbm.rgd->rd_bits[0].bi_bh); 2434 gfs2_rgrp_out(rbm.rgd, rbm.rgd->rd_bits[0].bi_bh->b_data); 2435 2436 gfs2_statfs_change(sdp, 0, -(s64)*nblocks, dinode ? 1 : 0); 2437 if (dinode) 2438 gfs2_trans_remove_revoke(sdp, block, *nblocks); 2439 2440 gfs2_quota_change(ip, *nblocks, ip->i_inode.i_uid, ip->i_inode.i_gid); 2441 2442 rbm.rgd->rd_free_clone -= *nblocks; 2443 trace_gfs2_block_alloc(ip, rbm.rgd, block, *nblocks, 2444 dinode ? GFS2_BLKST_DINODE : GFS2_BLKST_USED); 2445 *bn = block; 2446 return 0; 2447 2448 rgrp_error: 2449 gfs2_rgrp_error(rbm.rgd); 2450 return -EIO; 2451 } 2452 2453 /** 2454 * __gfs2_free_blocks - free a contiguous run of block(s) 2455 * @ip: the inode these blocks are being freed from 2456 * @rgd: the resource group the blocks are in 2457 * @bstart: first block of a run of contiguous blocks 2458 * @blen: the length of the block run 2459 * @meta: 1 if the blocks represent metadata 2460 * 2461 */ 2462 2463 void __gfs2_free_blocks(struct gfs2_inode *ip, struct gfs2_rgrpd *rgd, 2464 u64 bstart, u32 blen, int meta) 2465 { 2466 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 2467 2468 rgblk_free(sdp, rgd, bstart, blen, GFS2_BLKST_FREE); 2469 trace_gfs2_block_alloc(ip, rgd, bstart, blen, GFS2_BLKST_FREE); 2470 rgd->rd_free += blen; 2471 rgd->rd_flags &= ~GFS2_RGF_TRIMMED; 2472 gfs2_trans_add_meta(rgd->rd_gl, rgd->rd_bits[0].bi_bh); 2473 gfs2_rgrp_out(rgd, rgd->rd_bits[0].bi_bh->b_data); 2474 2475 /* Directories keep their data in the metadata address space */ 2476 if (meta || ip->i_depth) 2477 gfs2_meta_wipe(ip, bstart, blen); 2478 } 2479 2480 /** 2481 * gfs2_free_meta - free a contiguous run of data block(s) 2482 * @ip: the inode these blocks are being freed from 2483 * @rgd: the resource group the blocks are in 2484 * @bstart: first block of a run of contiguous blocks 2485 * @blen: the length of the block run 2486 * 2487 */ 2488 2489 void gfs2_free_meta(struct gfs2_inode *ip, struct gfs2_rgrpd *rgd, 2490 u64 bstart, u32 blen) 2491 { 2492 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 2493 2494 __gfs2_free_blocks(ip, rgd, bstart, blen, 1); 2495 gfs2_statfs_change(sdp, 0, +blen, 0); 2496 gfs2_quota_change(ip, -(s64)blen, ip->i_inode.i_uid, ip->i_inode.i_gid); 2497 } 2498 2499 void gfs2_unlink_di(struct inode *inode) 2500 { 2501 struct gfs2_inode *ip = GFS2_I(inode); 2502 struct gfs2_sbd *sdp = GFS2_SB(inode); 2503 struct gfs2_rgrpd *rgd; 2504 u64 blkno = ip->i_no_addr; 2505 2506 rgd = gfs2_blk2rgrpd(sdp, blkno, true); 2507 if (!rgd) 2508 return; 2509 rgblk_free(sdp, rgd, blkno, 1, GFS2_BLKST_UNLINKED); 2510 trace_gfs2_block_alloc(ip, rgd, blkno, 1, GFS2_BLKST_UNLINKED); 2511 gfs2_trans_add_meta(rgd->rd_gl, rgd->rd_bits[0].bi_bh); 2512 gfs2_rgrp_out(rgd, rgd->rd_bits[0].bi_bh->b_data); 2513 be32_add_cpu(&rgd->rd_rgl->rl_unlinked, 1); 2514 } 2515 2516 void gfs2_free_di(struct gfs2_rgrpd *rgd, struct gfs2_inode *ip) 2517 { 2518 struct gfs2_sbd *sdp = rgd->rd_sbd; 2519 2520 rgblk_free(sdp, rgd, ip->i_no_addr, 1, GFS2_BLKST_FREE); 2521 if (!rgd->rd_dinodes) 2522 gfs2_consist_rgrpd(rgd); 2523 rgd->rd_dinodes--; 2524 rgd->rd_free++; 2525 2526 gfs2_trans_add_meta(rgd->rd_gl, rgd->rd_bits[0].bi_bh); 2527 gfs2_rgrp_out(rgd, rgd->rd_bits[0].bi_bh->b_data); 2528 be32_add_cpu(&rgd->rd_rgl->rl_unlinked, -1); 2529 2530 gfs2_statfs_change(sdp, 0, +1, -1); 2531 trace_gfs2_block_alloc(ip, rgd, ip->i_no_addr, 1, GFS2_BLKST_FREE); 2532 gfs2_quota_change(ip, -1, ip->i_inode.i_uid, ip->i_inode.i_gid); 2533 gfs2_meta_wipe(ip, ip->i_no_addr, 1); 2534 } 2535 2536 /** 2537 * gfs2_check_blk_type - Check the type of a block 2538 * @sdp: The superblock 2539 * @no_addr: The block number to check 2540 * @type: The block type we are looking for 2541 * 2542 * Returns: 0 if the block type matches the expected type 2543 * -ESTALE if it doesn't match 2544 * or -ve errno if something went wrong while checking 2545 */ 2546 2547 int gfs2_check_blk_type(struct gfs2_sbd *sdp, u64 no_addr, unsigned int type) 2548 { 2549 struct gfs2_rgrpd *rgd; 2550 struct gfs2_holder rgd_gh; 2551 struct gfs2_rbm rbm; 2552 int error = -EINVAL; 2553 2554 rgd = gfs2_blk2rgrpd(sdp, no_addr, 1); 2555 if (!rgd) 2556 goto fail; 2557 2558 error = gfs2_glock_nq_init(rgd->rd_gl, LM_ST_SHARED, 0, &rgd_gh); 2559 if (error) 2560 goto fail; 2561 2562 rbm.rgd = rgd; 2563 error = gfs2_rbm_from_block(&rbm, no_addr); 2564 if (WARN_ON_ONCE(error)) 2565 goto fail; 2566 2567 if (gfs2_testbit(&rbm, false) != type) 2568 error = -ESTALE; 2569 2570 gfs2_glock_dq_uninit(&rgd_gh); 2571 fail: 2572 return error; 2573 } 2574 2575 /** 2576 * gfs2_rlist_add - add a RG to a list of RGs 2577 * @ip: the inode 2578 * @rlist: the list of resource groups 2579 * @block: the block 2580 * 2581 * Figure out what RG a block belongs to and add that RG to the list 2582 * 2583 * FIXME: Don't use NOFAIL 2584 * 2585 */ 2586 2587 void gfs2_rlist_add(struct gfs2_inode *ip, struct gfs2_rgrp_list *rlist, 2588 u64 block) 2589 { 2590 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); 2591 struct gfs2_rgrpd *rgd; 2592 struct gfs2_rgrpd **tmp; 2593 unsigned int new_space; 2594 unsigned int x; 2595 2596 if (gfs2_assert_warn(sdp, !rlist->rl_ghs)) 2597 return; 2598 2599 /* 2600 * The resource group last accessed is kept in the last position. 2601 */ 2602 2603 if (rlist->rl_rgrps) { 2604 rgd = rlist->rl_rgd[rlist->rl_rgrps - 1]; 2605 if (rgrp_contains_block(rgd, block)) 2606 return; 2607 rgd = gfs2_blk2rgrpd(sdp, block, 1); 2608 } else { 2609 rgd = ip->i_res.rs_rbm.rgd; 2610 if (!rgd || !rgrp_contains_block(rgd, block)) 2611 rgd = gfs2_blk2rgrpd(sdp, block, 1); 2612 } 2613 2614 if (!rgd) { 2615 fs_err(sdp, "rlist_add: no rgrp for block %llu\n", 2616 (unsigned long long)block); 2617 return; 2618 } 2619 2620 for (x = 0; x < rlist->rl_rgrps; x++) { 2621 if (rlist->rl_rgd[x] == rgd) { 2622 swap(rlist->rl_rgd[x], 2623 rlist->rl_rgd[rlist->rl_rgrps - 1]); 2624 return; 2625 } 2626 } 2627 2628 if (rlist->rl_rgrps == rlist->rl_space) { 2629 new_space = rlist->rl_space + 10; 2630 2631 tmp = kcalloc(new_space, sizeof(struct gfs2_rgrpd *), 2632 GFP_NOFS | __GFP_NOFAIL); 2633 2634 if (rlist->rl_rgd) { 2635 memcpy(tmp, rlist->rl_rgd, 2636 rlist->rl_space * sizeof(struct gfs2_rgrpd *)); 2637 kfree(rlist->rl_rgd); 2638 } 2639 2640 rlist->rl_space = new_space; 2641 rlist->rl_rgd = tmp; 2642 } 2643 2644 rlist->rl_rgd[rlist->rl_rgrps++] = rgd; 2645 } 2646 2647 /** 2648 * gfs2_rlist_alloc - all RGs have been added to the rlist, now allocate 2649 * and initialize an array of glock holders for them 2650 * @rlist: the list of resource groups 2651 * 2652 * FIXME: Don't use NOFAIL 2653 * 2654 */ 2655 2656 void gfs2_rlist_alloc(struct gfs2_rgrp_list *rlist) 2657 { 2658 unsigned int x; 2659 2660 rlist->rl_ghs = kmalloc_array(rlist->rl_rgrps, 2661 sizeof(struct gfs2_holder), 2662 GFP_NOFS | __GFP_NOFAIL); 2663 for (x = 0; x < rlist->rl_rgrps; x++) 2664 gfs2_holder_init(rlist->rl_rgd[x]->rd_gl, 2665 LM_ST_EXCLUSIVE, 0, 2666 &rlist->rl_ghs[x]); 2667 } 2668 2669 /** 2670 * gfs2_rlist_free - free a resource group list 2671 * @rlist: the list of resource groups 2672 * 2673 */ 2674 2675 void gfs2_rlist_free(struct gfs2_rgrp_list *rlist) 2676 { 2677 unsigned int x; 2678 2679 kfree(rlist->rl_rgd); 2680 2681 if (rlist->rl_ghs) { 2682 for (x = 0; x < rlist->rl_rgrps; x++) 2683 gfs2_holder_uninit(&rlist->rl_ghs[x]); 2684 kfree(rlist->rl_ghs); 2685 rlist->rl_ghs = NULL; 2686 } 2687 } 2688 2689