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 struct gfs2_rbm {
40 struct gfs2_rgrpd *rgd;
41 u32 offset; /* The offset is bitmap relative */
42 int bii; /* Bitmap index */
43 };
44
45 static inline struct gfs2_bitmap *rbm_bi(const struct gfs2_rbm *rbm)
46 {
47 return rbm->rgd->rd_bits + rbm->bii;
48 }
49
50 static inline u64 gfs2_rbm_to_block(const struct gfs2_rbm *rbm)
51 {
52 BUG_ON(rbm->offset >= rbm->rgd->rd_data);
53 return rbm->rgd->rd_data0 + (rbm_bi(rbm)->bi_start * GFS2_NBBY) +
54 rbm->offset;
55 }
56
57 /*
58 * These routines are used by the resource group routines (rgrp.c)
59 * to keep track of block allocation. Each block is represented by two
60 * bits. So, each byte represents GFS2_NBBY (i.e. 4) blocks.
61 *
62 * 0 = Free
63 * 1 = Used (not metadata)
64 * 2 = Unlinked (still in use) inode
65 * 3 = Used (metadata)
66 */
67
68 struct gfs2_extent {
69 struct gfs2_rbm rbm;
70 u32 len;
71 };
72
73 static const char valid_change[16] = {
74 /* current */
75 /* n */ 0, 1, 1, 1,
76 /* e */ 1, 0, 0, 0,
77 /* w */ 0, 0, 0, 1,
78 1, 0, 0, 0
79 };
80
81 static int gfs2_rbm_find(struct gfs2_rbm *rbm, u8 state, u32 *minext,
82 struct gfs2_blkreserv *rs, bool nowrap);
83
84
85 /**
86 * gfs2_setbit - Set a bit in the bitmaps
87 * @rbm: The position of the bit to set
88 * @do_clone: Also set the clone bitmap, if it exists
89 * @new_state: the new state of the block
90 *
91 */
92
93 static inline void gfs2_setbit(const struct gfs2_rbm *rbm, bool do_clone,
94 unsigned char new_state)
95 {
96 unsigned char *byte1, *byte2, *end, cur_state;
97 struct gfs2_bitmap *bi = rbm_bi(rbm);
98 unsigned int buflen = bi->bi_bytes;
99 const unsigned int bit = (rbm->offset % GFS2_NBBY) * GFS2_BIT_SIZE;
100
101 byte1 = bi->bi_bh->b_data + bi->bi_offset + (rbm->offset / GFS2_NBBY);
102 end = bi->bi_bh->b_data + bi->bi_offset + buflen;
103
104 BUG_ON(byte1 >= end);
105
106 cur_state = (*byte1 >> bit) & GFS2_BIT_MASK;
107
108 if (unlikely(!valid_change[new_state * 4 + cur_state])) {
109 struct gfs2_sbd *sdp = rbm->rgd->rd_sbd;
110
111 fs_warn(sdp, "buf_blk = 0x%x old_state=%d, new_state=%d\n",
112 rbm->offset, cur_state, new_state);
113 fs_warn(sdp, "rgrp=0x%llx bi_start=0x%x biblk: 0x%llx\n",
114 (unsigned long long)rbm->rgd->rd_addr, bi->bi_start,
115 (unsigned long long)bi->bi_bh->b_blocknr);
116 fs_warn(sdp, "bi_offset=0x%x bi_bytes=0x%x block=0x%llx\n",
117 bi->bi_offset, bi->bi_bytes,
118 (unsigned long long)gfs2_rbm_to_block(rbm));
119 dump_stack();
120 gfs2_consist_rgrpd(rbm->rgd);
121 return;
122 }
123 *byte1 ^= (cur_state ^ new_state) << bit;
124
125 if (do_clone && bi->bi_clone) {
126 byte2 = bi->bi_clone + bi->bi_offset + (rbm->offset / GFS2_NBBY);
127 cur_state = (*byte2 >> bit) & GFS2_BIT_MASK;
128 *byte2 ^= (cur_state ^ new_state) << bit;
129 }
130 }
131
132 /**
133 * gfs2_testbit - test a bit in the bitmaps
134 * @rbm: The bit to test
135 * @use_clone: If true, test the clone bitmap, not the official bitmap.
136 *
137 * Some callers like gfs2_unaligned_extlen need to test the clone bitmaps,
138 * not the "real" bitmaps, to avoid allocating recently freed blocks.
139 *
140 * Returns: The two bit block state of the requested bit
141 */
142
143 static inline u8 gfs2_testbit(const struct gfs2_rbm *rbm, bool use_clone)
144 {
145 struct gfs2_bitmap *bi = rbm_bi(rbm);
146 const u8 *buffer;
147 const u8 *byte;
148 unsigned int bit;
149
150 if (use_clone && bi->bi_clone)
151 buffer = bi->bi_clone;
152 else
153 buffer = bi->bi_bh->b_data;
154 buffer += bi->bi_offset;
155 byte = buffer + (rbm->offset / GFS2_NBBY);
156 bit = (rbm->offset % GFS2_NBBY) * GFS2_BIT_SIZE;
157
158 return (*byte >> bit) & GFS2_BIT_MASK;
159 }
160
161 /**
162 * gfs2_bit_search - search bitmap for a state
163 * @ptr: Pointer to bitmap data
164 * @mask: Mask to use (normally 0x55555.... but adjusted for search start)
165 * @state: The state we are searching for
166 *
167 * We xor the bitmap data with a pattern which is the bitwise opposite
168 * of what we are looking for. This gives rise to a pattern of ones
169 * wherever there is a match. Since we have two bits per entry, we
170 * take this pattern, shift it down by one place and then and it with
171 * the original. All the even bit positions (0,2,4, etc) then represent
172 * successful matches, so we mask with 0x55555..... to remove the unwanted
173 * odd bit positions.
174 *
175 * This allows searching of a whole u64 at once (32 blocks) with a
176 * single test (on 64 bit arches).
177 */
178
179 static inline u64 gfs2_bit_search(const __le64 *ptr, u64 mask, u8 state)
180 {
181 u64 tmp;
182 static const u64 search[] = {
183 [0] = 0xffffffffffffffffULL,
184 [1] = 0xaaaaaaaaaaaaaaaaULL,
185 [2] = 0x5555555555555555ULL,
186 [3] = 0x0000000000000000ULL,
187 };
188 tmp = le64_to_cpu(*ptr) ^ search[state];
189 tmp &= (tmp >> 1);
190 tmp &= mask;
191 return tmp;
192 }
193
194 /**
195 * rs_cmp - multi-block reservation range compare
196 * @start: start of the new reservation
197 * @len: number of blocks in the new reservation
198 * @rs: existing reservation to compare against
199 *
200 * returns: 1 if the block range is beyond the reach of the reservation
201 * -1 if the block range is before the start of the reservation
202 * 0 if the block range overlaps with the reservation
203 */
204 static inline int rs_cmp(u64 start, u32 len, struct gfs2_blkreserv *rs)
205 {
206 if (start >= rs->rs_start + rs->rs_requested)
207 return 1;
208 if (rs->rs_start >= start + len)
209 return -1;
210 return 0;
211 }
212
213 /**
214 * gfs2_bitfit - Search an rgrp's bitmap buffer to find a bit-pair representing
215 * a block in a given allocation state.
216 * @buf: the buffer that holds the bitmaps
217 * @len: the length (in bytes) of the buffer
218 * @goal: start search at this block's bit-pair (within @buffer)
219 * @state: GFS2_BLKST_XXX the state of the block we're looking for.
220 *
221 * Scope of @goal and returned block number is only within this bitmap buffer,
222 * not entire rgrp or filesystem. @buffer will be offset from the actual
223 * beginning of a bitmap block buffer, skipping any header structures, but
224 * headers are always a multiple of 64 bits long so that the buffer is
225 * always aligned to a 64 bit boundary.
226 *
227 * The size of the buffer is in bytes, but is it assumed that it is
228 * always ok to read a complete multiple of 64 bits at the end
229 * of the block in case the end is no aligned to a natural boundary.
230 *
231 * Return: the block number (bitmap buffer scope) that was found
232 */
233
234 static u32 gfs2_bitfit(const u8 *buf, const unsigned int len,
235 u32 goal, u8 state)
236 {
237 u32 spoint = (goal << 1) & ((8*sizeof(u64)) - 1);
238 const __le64 *ptr = ((__le64 *)buf) + (goal >> 5);
239 const __le64 *end = (__le64 *)(buf + ALIGN(len, sizeof(u64)));
240 u64 tmp;
241 u64 mask = 0x5555555555555555ULL;
242 u32 bit;
243
244 /* Mask off bits we don't care about at the start of the search */
245 mask <<= spoint;
246 tmp = gfs2_bit_search(ptr, mask, state);
247 ptr++;
248 while(tmp == 0 && ptr < end) {
249 tmp = gfs2_bit_search(ptr, 0x5555555555555555ULL, state);
250 ptr++;
251 }
252 /* Mask off any bits which are more than len bytes from the start */
253 if (ptr == end && (len & (sizeof(u64) - 1)))
254 tmp &= (((u64)~0) >> (64 - 8*(len & (sizeof(u64) - 1))));
255 /* Didn't find anything, so return */
256 if (tmp == 0)
257 return BFITNOENT;
258 ptr--;
259 bit = __ffs64(tmp);
260 bit /= 2; /* two bits per entry in the bitmap */
261 return (((const unsigned char *)ptr - buf) * GFS2_NBBY) + bit;
262 }
263
264 /**
265 * gfs2_rbm_from_block - Set the rbm based upon rgd and block number
266 * @rbm: The rbm with rgd already set correctly
267 * @block: The block number (filesystem relative)
268 *
269 * This sets the bi and offset members of an rbm based on a
270 * resource group and a filesystem relative block number. The
271 * resource group must be set in the rbm on entry, the bi and
272 * offset members will be set by this function.
273 *
274 * Returns: 0 on success, or an error code
275 */
276
277 static int gfs2_rbm_from_block(struct gfs2_rbm *rbm, u64 block)
278 {
279 if (!rgrp_contains_block(rbm->rgd, block))
280 return -E2BIG;
281 rbm->bii = 0;
282 rbm->offset = block - rbm->rgd->rd_data0;
283 /* Check if the block is within the first block */
284 if (rbm->offset < rbm_bi(rbm)->bi_blocks)
285 return 0;
286
287 /* Adjust for the size diff between gfs2_meta_header and gfs2_rgrp */
288 rbm->offset += (sizeof(struct gfs2_rgrp) -
289 sizeof(struct gfs2_meta_header)) * GFS2_NBBY;
290 rbm->bii = rbm->offset / rbm->rgd->rd_sbd->sd_blocks_per_bitmap;
291 rbm->offset -= rbm->bii * rbm->rgd->rd_sbd->sd_blocks_per_bitmap;
292 return 0;
293 }
294
295 /**
296 * gfs2_rbm_add - add a number of blocks to an rbm
297 * @rbm: The rbm with rgd already set correctly
298 * @blocks: The number of blocks to add to rpm
299 *
300 * This function takes an existing rbm structure and adds a number of blocks to
301 * it.
302 *
303 * Returns: True if the new rbm would point past the end of the rgrp.
304 */
305
306 static bool gfs2_rbm_add(struct gfs2_rbm *rbm, u32 blocks)
307 {
308 struct gfs2_rgrpd *rgd = rbm->rgd;
309 struct gfs2_bitmap *bi = rgd->rd_bits + rbm->bii;
310
311 if (rbm->offset + blocks < bi->bi_blocks) {
312 rbm->offset += blocks;
313 return false;
314 }
315 blocks -= bi->bi_blocks - rbm->offset;
316
317 for(;;) {
318 bi++;
319 if (bi == rgd->rd_bits + rgd->rd_length)
320 return true;
321 if (blocks < bi->bi_blocks) {
322 rbm->offset = blocks;
323 rbm->bii = bi - rgd->rd_bits;
324 return false;
325 }
326 blocks -= bi->bi_blocks;
327 }
328 }
329
330 /**
331 * gfs2_unaligned_extlen - Look for free blocks which are not byte aligned
332 * @rbm: Position to search (value/result)
333 * @n_unaligned: Number of unaligned blocks to check
334 * @len: Decremented for each block found (terminate on zero)
335 *
336 * Returns: true if a non-free block is encountered or the end of the resource
337 * group is reached.
338 */
339
340 static bool gfs2_unaligned_extlen(struct gfs2_rbm *rbm, u32 n_unaligned, u32 *len)
341 {
342 u32 n;
343 u8 res;
344
345 for (n = 0; n < n_unaligned; n++) {
346 res = gfs2_testbit(rbm, true);
347 if (res != GFS2_BLKST_FREE)
348 return true;
349 (*len)--;
350 if (*len == 0)
351 return true;
352 if (gfs2_rbm_add(rbm, 1))
353 return true;
354 }
355
356 return false;
357 }
358
359 /**
360 * gfs2_free_extlen - Return extent length of free blocks
361 * @rrbm: Starting position
362 * @len: Max length to check
363 *
364 * Starting at the block specified by the rbm, see how many free blocks
365 * there are, not reading more than len blocks ahead. This can be done
366 * using memchr_inv when the blocks are byte aligned, but has to be done
367 * on a block by block basis in case of unaligned blocks. Also this
368 * function can cope with bitmap boundaries (although it must stop on
369 * a resource group boundary)
370 *
371 * Returns: Number of free blocks in the extent
372 */
373
374 static u32 gfs2_free_extlen(const struct gfs2_rbm *rrbm, u32 len)
375 {
376 struct gfs2_rbm rbm = *rrbm;
377 u32 n_unaligned = rbm.offset & 3;
378 u32 size = len;
379 u32 bytes;
380 u32 chunk_size;
381 u8 *ptr, *start, *end;
382 u64 block;
383 struct gfs2_bitmap *bi;
384
385 if (n_unaligned &&
386 gfs2_unaligned_extlen(&rbm, 4 - n_unaligned, &len))
387 goto out;
388
389 n_unaligned = len & 3;
390 /* Start is now byte aligned */
391 while (len > 3) {
392 bi = rbm_bi(&rbm);
393 start = bi->bi_bh->b_data;
394 if (bi->bi_clone)
395 start = bi->bi_clone;
396 start += bi->bi_offset;
397 end = start + bi->bi_bytes;
398 BUG_ON(rbm.offset & 3);
399 start += (rbm.offset / GFS2_NBBY);
400 bytes = min_t(u32, len / GFS2_NBBY, (end - start));
401 ptr = memchr_inv(start, 0, bytes);
402 chunk_size = ((ptr == NULL) ? bytes : (ptr - start));
403 chunk_size *= GFS2_NBBY;
404 BUG_ON(len < chunk_size);
405 len -= chunk_size;
406 block = gfs2_rbm_to_block(&rbm);
407 if (gfs2_rbm_from_block(&rbm, block + chunk_size)) {
408 n_unaligned = 0;
409 break;
410 }
411 if (ptr) {
412 n_unaligned = 3;
413 break;
414 }
415 n_unaligned = len & 3;
416 }
417
418 /* Deal with any bits left over at the end */
419 if (n_unaligned)
420 gfs2_unaligned_extlen(&rbm, n_unaligned, &len);
421 out:
422 return size - len;
423 }
424
425 /**
426 * gfs2_bitcount - count the number of bits in a certain state
427 * @rgd: the resource group descriptor
428 * @buffer: the buffer that holds the bitmaps
429 * @buflen: the length (in bytes) of the buffer
430 * @state: the state of the block we're looking for
431 *
432 * Returns: The number of bits
433 */
434
435 static u32 gfs2_bitcount(struct gfs2_rgrpd *rgd, const u8 *buffer,
436 unsigned int buflen, u8 state)
437 {
438 const u8 *byte = buffer;
439 const u8 *end = buffer + buflen;
440 const u8 state1 = state << 2;
441 const u8 state2 = state << 4;
442 const u8 state3 = state << 6;
443 u32 count = 0;
444
445 for (; byte < end; byte++) {
446 if (((*byte) & 0x03) == state)
447 count++;
448 if (((*byte) & 0x0C) == state1)
449 count++;
450 if (((*byte) & 0x30) == state2)
451 count++;
452 if (((*byte) & 0xC0) == state3)
453 count++;
454 }
455
456 return count;
457 }
458
459 /**
460 * gfs2_rgrp_verify - Verify that a resource group is consistent
461 * @rgd: the rgrp
462 *
463 */
464
465 void gfs2_rgrp_verify(struct gfs2_rgrpd *rgd)
466 {
467 struct gfs2_sbd *sdp = rgd->rd_sbd;
468 struct gfs2_bitmap *bi = NULL;
469 u32 length = rgd->rd_length;
470 u32 count[4], tmp;
471 int buf, x;
472
473 memset(count, 0, 4 * sizeof(u32));
474
475 /* Count # blocks in each of 4 possible allocation states */
476 for (buf = 0; buf < length; buf++) {
477 bi = rgd->rd_bits + buf;
478 for (x = 0; x < 4; x++)
479 count[x] += gfs2_bitcount(rgd,
480 bi->bi_bh->b_data +
481 bi->bi_offset,
482 bi->bi_bytes, x);
483 }
484
485 if (count[0] != rgd->rd_free) {
486 gfs2_lm(sdp, "free data mismatch: %u != %u\n",
487 count[0], rgd->rd_free);
488 gfs2_consist_rgrpd(rgd);
489 return;
490 }
491
492 tmp = rgd->rd_data - rgd->rd_free - rgd->rd_dinodes;
493 if (count[1] != tmp) {
494 gfs2_lm(sdp, "used data mismatch: %u != %u\n",
495 count[1], tmp);
496 gfs2_consist_rgrpd(rgd);
497 return;
498 }
499
500 if (count[2] + count[3] != rgd->rd_dinodes) {
501 gfs2_lm(sdp, "used metadata mismatch: %u != %u\n",
502 count[2] + count[3], rgd->rd_dinodes);
503 gfs2_consist_rgrpd(rgd);
504 return;
505 }
506 }
507
508 /**
509 * gfs2_blk2rgrpd - Find resource group for a given data/meta block number
510 * @sdp: The GFS2 superblock
511 * @blk: The data block number
512 * @exact: True if this needs to be an exact match
513 *
514 * The @exact argument should be set to true by most callers. The exception
515 * is when we need to match blocks which are not represented by the rgrp
516 * bitmap, but which are part of the rgrp (i.e. padding blocks) which are
517 * there for alignment purposes. Another way of looking at it is that @exact
518 * matches only valid data/metadata blocks, but with @exact false, it will
519 * match any block within the extent of the rgrp.
520 *
521 * Returns: The resource group, or NULL if not found
522 */
523
524 struct gfs2_rgrpd *gfs2_blk2rgrpd(struct gfs2_sbd *sdp, u64 blk, bool exact)
525 {
526 struct rb_node *n, *next;
527 struct gfs2_rgrpd *cur;
528
529 spin_lock(&sdp->sd_rindex_spin);
530 n = sdp->sd_rindex_tree.rb_node;
531 while (n) {
532 cur = rb_entry(n, struct gfs2_rgrpd, rd_node);
533 next = NULL;
534 if (blk < cur->rd_addr)
535 next = n->rb_left;
536 else if (blk >= cur->rd_data0 + cur->rd_data)
537 next = n->rb_right;
538 if (next == NULL) {
539 spin_unlock(&sdp->sd_rindex_spin);
540 if (exact) {
541 if (blk < cur->rd_addr)
542 return NULL;
543 if (blk >= cur->rd_data0 + cur->rd_data)
544 return NULL;
545 }
546 return cur;
547 }
548 n = next;
549 }
550 spin_unlock(&sdp->sd_rindex_spin);
551
552 return NULL;
553 }
554
555 /**
556 * gfs2_rgrpd_get_first - get the first Resource Group in the filesystem
557 * @sdp: The GFS2 superblock
558 *
559 * Returns: The first rgrp in the filesystem
560 */
561
562 struct gfs2_rgrpd *gfs2_rgrpd_get_first(struct gfs2_sbd *sdp)
563 {
564 const struct rb_node *n;
565 struct gfs2_rgrpd *rgd;
566
567 spin_lock(&sdp->sd_rindex_spin);
568 n = rb_first(&sdp->sd_rindex_tree);
569 rgd = rb_entry(n, struct gfs2_rgrpd, rd_node);
570 spin_unlock(&sdp->sd_rindex_spin);
571
572 return rgd;
573 }
574
575 /**
576 * gfs2_rgrpd_get_next - get the next RG
577 * @rgd: the resource group descriptor
578 *
579 * Returns: The next rgrp
580 */
581
582 struct gfs2_rgrpd *gfs2_rgrpd_get_next(struct gfs2_rgrpd *rgd)
583 {
584 struct gfs2_sbd *sdp = rgd->rd_sbd;
585 const struct rb_node *n;
586
587 spin_lock(&sdp->sd_rindex_spin);
588 n = rb_next(&rgd->rd_node);
589 if (n == NULL)
590 n = rb_first(&sdp->sd_rindex_tree);
591
592 if (unlikely(&rgd->rd_node == n)) {
593 spin_unlock(&sdp->sd_rindex_spin);
594 return NULL;
595 }
596 rgd = rb_entry(n, struct gfs2_rgrpd, rd_node);
597 spin_unlock(&sdp->sd_rindex_spin);
598 return rgd;
599 }
600
601 void check_and_update_goal(struct gfs2_inode *ip)
602 {
603 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
604 if (!ip->i_goal || gfs2_blk2rgrpd(sdp, ip->i_goal, 1) == NULL)
605 ip->i_goal = ip->i_no_addr;
606 }
607
608 void gfs2_free_clones(struct gfs2_rgrpd *rgd)
609 {
610 int x;
611
612 for (x = 0; x < rgd->rd_length; x++) {
613 struct gfs2_bitmap *bi = rgd->rd_bits + x;
614 kfree(bi->bi_clone);
615 bi->bi_clone = NULL;
616 }
617 }
618
619 static void dump_rs(struct seq_file *seq, const struct gfs2_blkreserv *rs,
620 const char *fs_id_buf)
621 {
622 struct gfs2_inode *ip = container_of(rs, struct gfs2_inode, i_res);
623
624 gfs2_print_dbg(seq, "%s B: n:%llu s:%llu f:%u\n",
625 fs_id_buf,
626 (unsigned long long)ip->i_no_addr,
627 (unsigned long long)rs->rs_start,
628 rs->rs_requested);
629 }
630
631 /**
632 * __rs_deltree - remove a multi-block reservation from the rgd tree
633 * @rs: The reservation to remove
634 *
635 */
636 static void __rs_deltree(struct gfs2_blkreserv *rs)
637 {
638 struct gfs2_rgrpd *rgd;
639
640 if (!gfs2_rs_active(rs))
641 return;
642
643 rgd = rs->rs_rgd;
644 trace_gfs2_rs(rs, TRACE_RS_TREEDEL);
645 rb_erase(&rs->rs_node, &rgd->rd_rstree);
646 RB_CLEAR_NODE(&rs->rs_node);
647
648 if (rs->rs_requested) {
649 /* return requested blocks to the rgrp */
650 BUG_ON(rs->rs_rgd->rd_requested < rs->rs_requested);
651 rs->rs_rgd->rd_requested -= rs->rs_requested;
652
653 /* The rgrp extent failure point is likely not to increase;
654 it will only do so if the freed blocks are somehow
655 contiguous with a span of free blocks that follows. Still,
656 it will force the number to be recalculated later. */
657 rgd->rd_extfail_pt += rs->rs_requested;
658 rs->rs_requested = 0;
659 }
660 }
661
662 /**
663 * gfs2_rs_deltree - remove a multi-block reservation from the rgd tree
664 * @rs: The reservation to remove
665 *
666 */
667 void gfs2_rs_deltree(struct gfs2_blkreserv *rs)
668 {
669 struct gfs2_rgrpd *rgd;
670
671 rgd = rs->rs_rgd;
672 if (rgd) {
673 spin_lock(&rgd->rd_rsspin);
674 __rs_deltree(rs);
675 BUG_ON(rs->rs_requested);
676 spin_unlock(&rgd->rd_rsspin);
677 }
678 }
679
680 /**
681 * gfs2_rs_delete - delete a multi-block reservation
682 * @ip: The inode for this reservation
683 *
684 */
685 void gfs2_rs_delete(struct gfs2_inode *ip)
686 {
687 struct inode *inode = &ip->i_inode;
688
689 down_write(&ip->i_rw_mutex);
690 if (atomic_read(&inode->i_writecount) <= 1)
691 gfs2_rs_deltree(&ip->i_res);
692 up_write(&ip->i_rw_mutex);
693 }
694
695 /**
696 * return_all_reservations - return all reserved blocks back to the rgrp.
697 * @rgd: the rgrp that needs its space back
698 *
699 * We previously reserved a bunch of blocks for allocation. Now we need to
700 * give them back. This leave the reservation structures in tact, but removes
701 * all of their corresponding "no-fly zones".
702 */
703 static void return_all_reservations(struct gfs2_rgrpd *rgd)
704 {
705 struct rb_node *n;
706 struct gfs2_blkreserv *rs;
707
708 spin_lock(&rgd->rd_rsspin);
709 while ((n = rb_first(&rgd->rd_rstree))) {
710 rs = rb_entry(n, struct gfs2_blkreserv, rs_node);
711 __rs_deltree(rs);
712 }
713 spin_unlock(&rgd->rd_rsspin);
714 }
715
716 void gfs2_clear_rgrpd(struct gfs2_sbd *sdp)
717 {
718 struct rb_node *n;
719 struct gfs2_rgrpd *rgd;
720 struct gfs2_glock *gl;
721
722 while ((n = rb_first(&sdp->sd_rindex_tree))) {
723 rgd = rb_entry(n, struct gfs2_rgrpd, rd_node);
724 gl = rgd->rd_gl;
725
726 rb_erase(n, &sdp->sd_rindex_tree);
727
728 if (gl) {
729 if (gl->gl_state != LM_ST_UNLOCKED) {
730 gfs2_glock_cb(gl, LM_ST_UNLOCKED);
731 flush_delayed_work(&gl->gl_work);
732 }
733 gfs2_rgrp_brelse(rgd);
734 glock_clear_object(gl, rgd);
735 gfs2_glock_put(gl);
736 }
737
738 gfs2_free_clones(rgd);
739 return_all_reservations(rgd);
740 kfree(rgd->rd_bits);
741 rgd->rd_bits = NULL;
742 kmem_cache_free(gfs2_rgrpd_cachep, rgd);
743 }
744 }
745
746 /**
747 * compute_bitstructs - Compute the bitmap sizes
748 * @rgd: The resource group descriptor
749 *
750 * Calculates bitmap descriptors, one for each block that contains bitmap data
751 *
752 * Returns: errno
753 */
754
755 static int compute_bitstructs(struct gfs2_rgrpd *rgd)
756 {
757 struct gfs2_sbd *sdp = rgd->rd_sbd;
758 struct gfs2_bitmap *bi;
759 u32 length = rgd->rd_length; /* # blocks in hdr & bitmap */
760 u32 bytes_left, bytes;
761 int x;
762
763 if (!length)
764 return -EINVAL;
765
766 rgd->rd_bits = kcalloc(length, sizeof(struct gfs2_bitmap), GFP_NOFS);
767 if (!rgd->rd_bits)
768 return -ENOMEM;
769
770 bytes_left = rgd->rd_bitbytes;
771
772 for (x = 0; x < length; x++) {
773 bi = rgd->rd_bits + x;
774
775 bi->bi_flags = 0;
776 /* small rgrp; bitmap stored completely in header block */
777 if (length == 1) {
778 bytes = bytes_left;
779 bi->bi_offset = sizeof(struct gfs2_rgrp);
780 bi->bi_start = 0;
781 bi->bi_bytes = bytes;
782 bi->bi_blocks = bytes * GFS2_NBBY;
783 /* header block */
784 } else if (x == 0) {
785 bytes = sdp->sd_sb.sb_bsize - sizeof(struct gfs2_rgrp);
786 bi->bi_offset = sizeof(struct gfs2_rgrp);
787 bi->bi_start = 0;
788 bi->bi_bytes = bytes;
789 bi->bi_blocks = bytes * GFS2_NBBY;
790 /* last block */
791 } else if (x + 1 == length) {
792 bytes = bytes_left;
793 bi->bi_offset = sizeof(struct gfs2_meta_header);
794 bi->bi_start = rgd->rd_bitbytes - bytes_left;
795 bi->bi_bytes = bytes;
796 bi->bi_blocks = bytes * GFS2_NBBY;
797 /* other blocks */
798 } else {
799 bytes = sdp->sd_sb.sb_bsize -
800 sizeof(struct gfs2_meta_header);
801 bi->bi_offset = sizeof(struct gfs2_meta_header);
802 bi->bi_start = rgd->rd_bitbytes - bytes_left;
803 bi->bi_bytes = bytes;
804 bi->bi_blocks = bytes * GFS2_NBBY;
805 }
806
807 bytes_left -= bytes;
808 }
809
810 if (bytes_left) {
811 gfs2_consist_rgrpd(rgd);
812 return -EIO;
813 }
814 bi = rgd->rd_bits + (length - 1);
815 if ((bi->bi_start + bi->bi_bytes) * GFS2_NBBY != rgd->rd_data) {
816 gfs2_lm(sdp,
817 "ri_addr=%llu "
818 "ri_length=%u "
819 "ri_data0=%llu "
820 "ri_data=%u "
821 "ri_bitbytes=%u "
822 "start=%u len=%u offset=%u\n",
823 (unsigned long long)rgd->rd_addr,
824 rgd->rd_length,
825 (unsigned long long)rgd->rd_data0,
826 rgd->rd_data,
827 rgd->rd_bitbytes,
828 bi->bi_start, bi->bi_bytes, bi->bi_offset);
829 gfs2_consist_rgrpd(rgd);
830 return -EIO;
831 }
832
833 return 0;
834 }
835
836 /**
837 * gfs2_ri_total - Total up the file system space, according to the rindex.
838 * @sdp: the filesystem
839 *
840 */
841 u64 gfs2_ri_total(struct gfs2_sbd *sdp)
842 {
843 u64 total_data = 0;
844 struct inode *inode = sdp->sd_rindex;
845 struct gfs2_inode *ip = GFS2_I(inode);
846 char buf[sizeof(struct gfs2_rindex)];
847 int error, rgrps;
848
849 for (rgrps = 0;; rgrps++) {
850 loff_t pos = rgrps * sizeof(struct gfs2_rindex);
851
852 if (pos + sizeof(struct gfs2_rindex) > i_size_read(inode))
853 break;
854 error = gfs2_internal_read(ip, buf, &pos,
855 sizeof(struct gfs2_rindex));
856 if (error != sizeof(struct gfs2_rindex))
857 break;
858 total_data += be32_to_cpu(((struct gfs2_rindex *)buf)->ri_data);
859 }
860 return total_data;
861 }
862
863 static int rgd_insert(struct gfs2_rgrpd *rgd)
864 {
865 struct gfs2_sbd *sdp = rgd->rd_sbd;
866 struct rb_node **newn = &sdp->sd_rindex_tree.rb_node, *parent = NULL;
867
868 /* Figure out where to put new node */
869 while (*newn) {
870 struct gfs2_rgrpd *cur = rb_entry(*newn, struct gfs2_rgrpd,
871 rd_node);
872
873 parent = *newn;
874 if (rgd->rd_addr < cur->rd_addr)
875 newn = &((*newn)->rb_left);
876 else if (rgd->rd_addr > cur->rd_addr)
877 newn = &((*newn)->rb_right);
878 else
879 return -EEXIST;
880 }
881
882 rb_link_node(&rgd->rd_node, parent, newn);
883 rb_insert_color(&rgd->rd_node, &sdp->sd_rindex_tree);
884 sdp->sd_rgrps++;
885 return 0;
886 }
887
888 /**
889 * read_rindex_entry - Pull in a new resource index entry from the disk
890 * @ip: Pointer to the rindex inode
891 *
892 * Returns: 0 on success, > 0 on EOF, error code otherwise
893 */
894
895 static int read_rindex_entry(struct gfs2_inode *ip)
896 {
897 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
898 loff_t pos = sdp->sd_rgrps * sizeof(struct gfs2_rindex);
899 struct gfs2_rindex buf;
900 int error;
901 struct gfs2_rgrpd *rgd;
902
903 if (pos >= i_size_read(&ip->i_inode))
904 return 1;
905
906 error = gfs2_internal_read(ip, (char *)&buf, &pos,
907 sizeof(struct gfs2_rindex));
908
909 if (error != sizeof(struct gfs2_rindex))
910 return (error == 0) ? 1 : error;
911
912 rgd = kmem_cache_zalloc(gfs2_rgrpd_cachep, GFP_NOFS);
913 error = -ENOMEM;
914 if (!rgd)
915 return error;
916
917 rgd->rd_sbd = sdp;
918 rgd->rd_addr = be64_to_cpu(buf.ri_addr);
919 rgd->rd_length = be32_to_cpu(buf.ri_length);
920 rgd->rd_data0 = be64_to_cpu(buf.ri_data0);
921 rgd->rd_data = be32_to_cpu(buf.ri_data);
922 rgd->rd_bitbytes = be32_to_cpu(buf.ri_bitbytes);
923 spin_lock_init(&rgd->rd_rsspin);
924 mutex_init(&rgd->rd_mutex);
925
926 error = gfs2_glock_get(sdp, rgd->rd_addr,
927 &gfs2_rgrp_glops, CREATE, &rgd->rd_gl);
928 if (error)
929 goto fail;
930
931 error = compute_bitstructs(rgd);
932 if (error)
933 goto fail_glock;
934
935 rgd->rd_rgl = (struct gfs2_rgrp_lvb *)rgd->rd_gl->gl_lksb.sb_lvbptr;
936 rgd->rd_flags &= ~GFS2_RDF_PREFERRED;
937 if (rgd->rd_data > sdp->sd_max_rg_data)
938 sdp->sd_max_rg_data = rgd->rd_data;
939 spin_lock(&sdp->sd_rindex_spin);
940 error = rgd_insert(rgd);
941 spin_unlock(&sdp->sd_rindex_spin);
942 if (!error) {
943 glock_set_object(rgd->rd_gl, rgd);
944 return 0;
945 }
946
947 error = 0; /* someone else read in the rgrp; free it and ignore it */
948 fail_glock:
949 gfs2_glock_put(rgd->rd_gl);
950
951 fail:
952 kfree(rgd->rd_bits);
953 rgd->rd_bits = NULL;
954 kmem_cache_free(gfs2_rgrpd_cachep, rgd);
955 return error;
956 }
957
958 /**
959 * set_rgrp_preferences - Run all the rgrps, selecting some we prefer to use
960 * @sdp: the GFS2 superblock
961 *
962 * The purpose of this function is to select a subset of the resource groups
963 * and mark them as PREFERRED. We do it in such a way that each node prefers
964 * to use a unique set of rgrps to minimize glock contention.
965 */
966 static void set_rgrp_preferences(struct gfs2_sbd *sdp)
967 {
968 struct gfs2_rgrpd *rgd, *first;
969 int i;
970
971 /* Skip an initial number of rgrps, based on this node's journal ID.
972 That should start each node out on its own set. */
973 rgd = gfs2_rgrpd_get_first(sdp);
974 for (i = 0; i < sdp->sd_lockstruct.ls_jid; i++)
975 rgd = gfs2_rgrpd_get_next(rgd);
976 first = rgd;
977
978 do {
979 rgd->rd_flags |= GFS2_RDF_PREFERRED;
980 for (i = 0; i < sdp->sd_journals; i++) {
981 rgd = gfs2_rgrpd_get_next(rgd);
982 if (!rgd || rgd == first)
983 break;
984 }
985 } while (rgd && rgd != first);
986 }
987
988 /**
989 * gfs2_ri_update - Pull in a new resource index from the disk
990 * @ip: pointer to the rindex inode
991 *
992 * Returns: 0 on successful update, error code otherwise
993 */
994
995 static int gfs2_ri_update(struct gfs2_inode *ip)
996 {
997 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
998 int error;
999
1000 do {
1001 error = read_rindex_entry(ip);
1002 } while (error == 0);
1003
1004 if (error < 0)
1005 return error;
1006
1007 if (RB_EMPTY_ROOT(&sdp->sd_rindex_tree)) {
1008 fs_err(sdp, "no resource groups found in the file system.\n");
1009 return -ENOENT;
1010 }
1011 set_rgrp_preferences(sdp);
1012
1013 sdp->sd_rindex_uptodate = 1;
1014 return 0;
1015 }
1016
1017 /**
1018 * gfs2_rindex_update - Update the rindex if required
1019 * @sdp: The GFS2 superblock
1020 *
1021 * We grab a lock on the rindex inode to make sure that it doesn't
1022 * change whilst we are performing an operation. We keep this lock
1023 * for quite long periods of time compared to other locks. This
1024 * doesn't matter, since it is shared and it is very, very rarely
1025 * accessed in the exclusive mode (i.e. only when expanding the filesystem).
1026 *
1027 * This makes sure that we're using the latest copy of the resource index
1028 * special file, which might have been updated if someone expanded the
1029 * filesystem (via gfs2_grow utility), which adds new resource groups.
1030 *
1031 * Returns: 0 on succeess, error code otherwise
1032 */
1033
1034 int gfs2_rindex_update(struct gfs2_sbd *sdp)
1035 {
1036 struct gfs2_inode *ip = GFS2_I(sdp->sd_rindex);
1037 struct gfs2_glock *gl = ip->i_gl;
1038 struct gfs2_holder ri_gh;
1039 int error = 0;
1040 int unlock_required = 0;
1041
1042 /* Read new copy from disk if we don't have the latest */
1043 if (!sdp->sd_rindex_uptodate) {
1044 if (!gfs2_glock_is_locked_by_me(gl)) {
1045 error = gfs2_glock_nq_init(gl, LM_ST_SHARED, 0, &ri_gh);
1046 if (error)
1047 return error;
1048 unlock_required = 1;
1049 }
1050 if (!sdp->sd_rindex_uptodate)
1051 error = gfs2_ri_update(ip);
1052 if (unlock_required)
1053 gfs2_glock_dq_uninit(&ri_gh);
1054 }
1055
1056 return error;
1057 }
1058
1059 static void gfs2_rgrp_in(struct gfs2_rgrpd *rgd, const void *buf)
1060 {
1061 const struct gfs2_rgrp *str = buf;
1062 u32 rg_flags;
1063
1064 rg_flags = be32_to_cpu(str->rg_flags);
1065 rg_flags &= ~GFS2_RDF_MASK;
1066 rgd->rd_flags &= GFS2_RDF_MASK;
1067 rgd->rd_flags |= rg_flags;
1068 rgd->rd_free = be32_to_cpu(str->rg_free);
1069 rgd->rd_dinodes = be32_to_cpu(str->rg_dinodes);
1070 rgd->rd_igeneration = be64_to_cpu(str->rg_igeneration);
1071 /* rd_data0, rd_data and rd_bitbytes already set from rindex */
1072 }
1073
1074 static void gfs2_rgrp_ondisk2lvb(struct gfs2_rgrp_lvb *rgl, const void *buf)
1075 {
1076 const struct gfs2_rgrp *str = buf;
1077
1078 rgl->rl_magic = cpu_to_be32(GFS2_MAGIC);
1079 rgl->rl_flags = str->rg_flags;
1080 rgl->rl_free = str->rg_free;
1081 rgl->rl_dinodes = str->rg_dinodes;
1082 rgl->rl_igeneration = str->rg_igeneration;
1083 rgl->__pad = 0UL;
1084 }
1085
1086 static void gfs2_rgrp_out(struct gfs2_rgrpd *rgd, void *buf)
1087 {
1088 struct gfs2_rgrpd *next = gfs2_rgrpd_get_next(rgd);
1089 struct gfs2_rgrp *str = buf;
1090 u32 crc;
1091
1092 str->rg_flags = cpu_to_be32(rgd->rd_flags & ~GFS2_RDF_MASK);
1093 str->rg_free = cpu_to_be32(rgd->rd_free);
1094 str->rg_dinodes = cpu_to_be32(rgd->rd_dinodes);
1095 if (next == NULL)
1096 str->rg_skip = 0;
1097 else if (next->rd_addr > rgd->rd_addr)
1098 str->rg_skip = cpu_to_be32(next->rd_addr - rgd->rd_addr);
1099 str->rg_igeneration = cpu_to_be64(rgd->rd_igeneration);
1100 str->rg_data0 = cpu_to_be64(rgd->rd_data0);
1101 str->rg_data = cpu_to_be32(rgd->rd_data);
1102 str->rg_bitbytes = cpu_to_be32(rgd->rd_bitbytes);
1103 str->rg_crc = 0;
1104 crc = gfs2_disk_hash(buf, sizeof(struct gfs2_rgrp));
1105 str->rg_crc = cpu_to_be32(crc);
1106
1107 memset(&str->rg_reserved, 0, sizeof(str->rg_reserved));
1108 gfs2_rgrp_ondisk2lvb(rgd->rd_rgl, buf);
1109 }
1110
1111 static int gfs2_rgrp_lvb_valid(struct gfs2_rgrpd *rgd)
1112 {
1113 struct gfs2_rgrp_lvb *rgl = rgd->rd_rgl;
1114 struct gfs2_rgrp *str = (struct gfs2_rgrp *)rgd->rd_bits[0].bi_bh->b_data;
1115 struct gfs2_sbd *sdp = rgd->rd_sbd;
1116 int valid = 1;
1117
1118 if (rgl->rl_flags != str->rg_flags) {
1119 fs_warn(sdp, "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 fs_warn(sdp, "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 fs_warn(sdp, "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 fs_warn(sdp, "GFS2: rgd: %llu lvb igen mismatch %llu/%llu",
1139 (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 static void rgrp_set_bitmap_flags(struct gfs2_rgrpd *rgd)
1172 {
1173 struct gfs2_bitmap *bi;
1174 int x;
1175
1176 if (rgd->rd_free) {
1177 for (x = 0; x < rgd->rd_length; x++) {
1178 bi = rgd->rd_bits + x;
1179 clear_bit(GBF_FULL, &bi->bi_flags);
1180 }
1181 } else {
1182 for (x = 0; x < rgd->rd_length; x++) {
1183 bi = rgd->rd_bits + x;
1184 set_bit(GBF_FULL, &bi->bi_flags);
1185 }
1186 }
1187 }
1188
1189 /**
1190 * gfs2_rgrp_go_instantiate - Read in a RG's header and bitmaps
1191 * @gl: the glock representing the rgrpd to read in
1192 *
1193 * Read in all of a Resource Group's header and bitmap blocks.
1194 * Caller must eventually call gfs2_rgrp_brelse() to free the bitmaps.
1195 *
1196 * Returns: errno
1197 */
1198
1199 int gfs2_rgrp_go_instantiate(struct gfs2_glock *gl)
1200 {
1201 struct gfs2_rgrpd *rgd = gl->gl_object;
1202 struct gfs2_sbd *sdp = rgd->rd_sbd;
1203 unsigned int length = rgd->rd_length;
1204 struct gfs2_bitmap *bi;
1205 unsigned int x, y;
1206 int error;
1207
1208 if (rgd->rd_bits[0].bi_bh != NULL)
1209 return 0;
1210
1211 for (x = 0; x < length; x++) {
1212 bi = rgd->rd_bits + x;
1213 error = gfs2_meta_read(gl, rgd->rd_addr + x, 0, 0, &bi->bi_bh);
1214 if (error)
1215 goto fail;
1216 }
1217
1218 for (y = length; y--;) {
1219 bi = rgd->rd_bits + y;
1220 error = gfs2_meta_wait(sdp, bi->bi_bh);
1221 if (error)
1222 goto fail;
1223 if (gfs2_metatype_check(sdp, bi->bi_bh, y ? GFS2_METATYPE_RB :
1224 GFS2_METATYPE_RG)) {
1225 error = -EIO;
1226 goto fail;
1227 }
1228 }
1229
1230 gfs2_rgrp_in(rgd, (rgd->rd_bits[0].bi_bh)->b_data);
1231 rgrp_set_bitmap_flags(rgd);
1232 rgd->rd_flags |= GFS2_RDF_CHECK;
1233 rgd->rd_free_clone = rgd->rd_free;
1234 GLOCK_BUG_ON(rgd->rd_gl, rgd->rd_reserved);
1235 /* max out the rgrp allocation failure point */
1236 rgd->rd_extfail_pt = rgd->rd_free;
1237 if (cpu_to_be32(GFS2_MAGIC) != rgd->rd_rgl->rl_magic) {
1238 rgd->rd_rgl->rl_unlinked = cpu_to_be32(count_unlinked(rgd));
1239 gfs2_rgrp_ondisk2lvb(rgd->rd_rgl,
1240 rgd->rd_bits[0].bi_bh->b_data);
1241 } else if (sdp->sd_args.ar_rgrplvb) {
1242 if (!gfs2_rgrp_lvb_valid(rgd)){
1243 gfs2_consist_rgrpd(rgd);
1244 error = -EIO;
1245 goto fail;
1246 }
1247 if (rgd->rd_rgl->rl_unlinked == 0)
1248 rgd->rd_flags &= ~GFS2_RDF_CHECK;
1249 }
1250 return 0;
1251
1252 fail:
1253 while (x--) {
1254 bi = rgd->rd_bits + x;
1255 brelse(bi->bi_bh);
1256 bi->bi_bh = NULL;
1257 gfs2_assert_warn(sdp, !bi->bi_clone);
1258 }
1259 return error;
1260 }
1261
1262 static int update_rgrp_lvb(struct gfs2_rgrpd *rgd, struct gfs2_holder *gh)
1263 {
1264 u32 rl_flags;
1265
1266 if (!test_bit(GLF_INSTANTIATE_NEEDED, &gh->gh_gl->gl_flags))
1267 return 0;
1268
1269 if (cpu_to_be32(GFS2_MAGIC) != rgd->rd_rgl->rl_magic)
1270 return gfs2_instantiate(gh);
1271
1272 rl_flags = be32_to_cpu(rgd->rd_rgl->rl_flags);
1273 rl_flags &= ~GFS2_RDF_MASK;
1274 rgd->rd_flags &= GFS2_RDF_MASK;
1275 rgd->rd_flags |= (rl_flags | GFS2_RDF_CHECK);
1276 if (rgd->rd_rgl->rl_unlinked == 0)
1277 rgd->rd_flags &= ~GFS2_RDF_CHECK;
1278 rgd->rd_free = be32_to_cpu(rgd->rd_rgl->rl_free);
1279 rgrp_set_bitmap_flags(rgd);
1280 rgd->rd_free_clone = rgd->rd_free;
1281 GLOCK_BUG_ON(rgd->rd_gl, rgd->rd_reserved);
1282 /* max out the rgrp allocation failure point */
1283 rgd->rd_extfail_pt = rgd->rd_free;
1284 rgd->rd_dinodes = be32_to_cpu(rgd->rd_rgl->rl_dinodes);
1285 rgd->rd_igeneration = be64_to_cpu(rgd->rd_rgl->rl_igeneration);
1286 return 0;
1287 }
1288
1289 /**
1290 * gfs2_rgrp_brelse - Release RG bitmaps read in with gfs2_rgrp_bh_get()
1291 * @rgd: The resource group
1292 *
1293 */
1294
1295 void gfs2_rgrp_brelse(struct gfs2_rgrpd *rgd)
1296 {
1297 int x, length = rgd->rd_length;
1298
1299 for (x = 0; x < length; x++) {
1300 struct gfs2_bitmap *bi = rgd->rd_bits + x;
1301 if (bi->bi_bh) {
1302 brelse(bi->bi_bh);
1303 bi->bi_bh = NULL;
1304 }
1305 }
1306 set_bit(GLF_INSTANTIATE_NEEDED, &rgd->rd_gl->gl_flags);
1307 }
1308
1309 int gfs2_rgrp_send_discards(struct gfs2_sbd *sdp, u64 offset,
1310 struct buffer_head *bh,
1311 const struct gfs2_bitmap *bi, unsigned minlen, u64 *ptrimmed)
1312 {
1313 struct super_block *sb = sdp->sd_vfs;
1314 u64 blk;
1315 sector_t start = 0;
1316 sector_t nr_blks = 0;
1317 int rv = -EIO;
1318 unsigned int x;
1319 u32 trimmed = 0;
1320 u8 diff;
1321
1322 for (x = 0; x < bi->bi_bytes; x++) {
1323 const u8 *clone = bi->bi_clone ? bi->bi_clone : bi->bi_bh->b_data;
1324 clone += bi->bi_offset;
1325 clone += x;
1326 if (bh) {
1327 const u8 *orig = bh->b_data + bi->bi_offset + x;
1328 diff = ~(*orig | (*orig >> 1)) & (*clone | (*clone >> 1));
1329 } else {
1330 diff = ~(*clone | (*clone >> 1));
1331 }
1332 diff &= 0x55;
1333 if (diff == 0)
1334 continue;
1335 blk = offset + ((bi->bi_start + x) * GFS2_NBBY);
1336 while(diff) {
1337 if (diff & 1) {
1338 if (nr_blks == 0)
1339 goto start_new_extent;
1340 if ((start + nr_blks) != blk) {
1341 if (nr_blks >= minlen) {
1342 rv = sb_issue_discard(sb,
1343 start, nr_blks,
1344 GFP_NOFS, 0);
1345 if (rv)
1346 goto fail;
1347 trimmed += nr_blks;
1348 }
1349 nr_blks = 0;
1350 start_new_extent:
1351 start = blk;
1352 }
1353 nr_blks++;
1354 }
1355 diff >>= 2;
1356 blk++;
1357 }
1358 }
1359 if (nr_blks >= minlen) {
1360 rv = sb_issue_discard(sb, start, nr_blks, GFP_NOFS, 0);
1361 if (rv)
1362 goto fail;
1363 trimmed += nr_blks;
1364 }
1365 if (ptrimmed)
1366 *ptrimmed = trimmed;
1367 return 0;
1368
1369 fail:
1370 if (sdp->sd_args.ar_discard)
1371 fs_warn(sdp, "error %d on discard request, turning discards off for this filesystem\n", rv);
1372 sdp->sd_args.ar_discard = 0;
1373 return rv;
1374 }
1375
1376 /**
1377 * gfs2_fitrim - Generate discard requests for unused bits of the filesystem
1378 * @filp: Any file on the filesystem
1379 * @argp: Pointer to the arguments (also used to pass result)
1380 *
1381 * Returns: 0 on success, otherwise error code
1382 */
1383
1384 int gfs2_fitrim(struct file *filp, void __user *argp)
1385 {
1386 struct inode *inode = file_inode(filp);
1387 struct gfs2_sbd *sdp = GFS2_SB(inode);
1388 struct block_device *bdev = sdp->sd_vfs->s_bdev;
1389 struct buffer_head *bh;
1390 struct gfs2_rgrpd *rgd;
1391 struct gfs2_rgrpd *rgd_end;
1392 struct gfs2_holder gh;
1393 struct fstrim_range r;
1394 int ret = 0;
1395 u64 amt;
1396 u64 trimmed = 0;
1397 u64 start, end, minlen;
1398 unsigned int x;
1399 unsigned bs_shift = sdp->sd_sb.sb_bsize_shift;
1400
1401 if (!capable(CAP_SYS_ADMIN))
1402 return -EPERM;
1403
1404 if (!test_bit(SDF_JOURNAL_LIVE, &sdp->sd_flags))
1405 return -EROFS;
1406
1407 if (!bdev_max_discard_sectors(bdev))
1408 return -EOPNOTSUPP;
1409
1410 if (copy_from_user(&r, argp, sizeof(r)))
1411 return -EFAULT;
1412
1413 ret = gfs2_rindex_update(sdp);
1414 if (ret)
1415 return ret;
1416
1417 start = r.start >> bs_shift;
1418 end = start + (r.len >> bs_shift);
1419 minlen = max_t(u64, r.minlen, sdp->sd_sb.sb_bsize);
1420 minlen = max_t(u64, minlen, bdev_discard_granularity(bdev)) >> bs_shift;
1421
1422 if (end <= start || minlen > sdp->sd_max_rg_data)
1423 return -EINVAL;
1424
1425 rgd = gfs2_blk2rgrpd(sdp, start, 0);
1426 rgd_end = gfs2_blk2rgrpd(sdp, end, 0);
1427
1428 if ((gfs2_rgrpd_get_first(sdp) == gfs2_rgrpd_get_next(rgd_end))
1429 && (start > rgd_end->rd_data0 + rgd_end->rd_data))
1430 return -EINVAL; /* start is beyond the end of the fs */
1431
1432 while (1) {
1433
1434 ret = gfs2_glock_nq_init(rgd->rd_gl, LM_ST_EXCLUSIVE,
1435 LM_FLAG_NODE_SCOPE, &gh);
1436 if (ret)
1437 goto out;
1438
1439 if (!(rgd->rd_flags & GFS2_RGF_TRIMMED)) {
1440 /* Trim each bitmap in the rgrp */
1441 for (x = 0; x < rgd->rd_length; x++) {
1442 struct gfs2_bitmap *bi = rgd->rd_bits + x;
1443 rgrp_lock_local(rgd);
1444 ret = gfs2_rgrp_send_discards(sdp,
1445 rgd->rd_data0, NULL, bi, minlen,
1446 &amt);
1447 rgrp_unlock_local(rgd);
1448 if (ret) {
1449 gfs2_glock_dq_uninit(&gh);
1450 goto out;
1451 }
1452 trimmed += amt;
1453 }
1454
1455 /* Mark rgrp as having been trimmed */
1456 ret = gfs2_trans_begin(sdp, RES_RG_HDR, 0);
1457 if (ret == 0) {
1458 bh = rgd->rd_bits[0].bi_bh;
1459 rgrp_lock_local(rgd);
1460 rgd->rd_flags |= GFS2_RGF_TRIMMED;
1461 gfs2_trans_add_meta(rgd->rd_gl, bh);
1462 gfs2_rgrp_out(rgd, bh->b_data);
1463 rgrp_unlock_local(rgd);
1464 gfs2_trans_end(sdp);
1465 }
1466 }
1467 gfs2_glock_dq_uninit(&gh);
1468
1469 if (rgd == rgd_end)
1470 break;
1471
1472 rgd = gfs2_rgrpd_get_next(rgd);
1473 }
1474
1475 out:
1476 r.len = trimmed << bs_shift;
1477 if (copy_to_user(argp, &r, sizeof(r)))
1478 return -EFAULT;
1479
1480 return ret;
1481 }
1482
1483 /**
1484 * rs_insert - insert a new multi-block reservation into the rgrp's rb_tree
1485 * @ip: the inode structure
1486 *
1487 */
1488 static void rs_insert(struct gfs2_inode *ip)
1489 {
1490 struct rb_node **newn, *parent = NULL;
1491 int rc;
1492 struct gfs2_blkreserv *rs = &ip->i_res;
1493 struct gfs2_rgrpd *rgd = rs->rs_rgd;
1494
1495 BUG_ON(gfs2_rs_active(rs));
1496
1497 spin_lock(&rgd->rd_rsspin);
1498 newn = &rgd->rd_rstree.rb_node;
1499 while (*newn) {
1500 struct gfs2_blkreserv *cur =
1501 rb_entry(*newn, struct gfs2_blkreserv, rs_node);
1502
1503 parent = *newn;
1504 rc = rs_cmp(rs->rs_start, rs->rs_requested, cur);
1505 if (rc > 0)
1506 newn = &((*newn)->rb_right);
1507 else if (rc < 0)
1508 newn = &((*newn)->rb_left);
1509 else {
1510 spin_unlock(&rgd->rd_rsspin);
1511 WARN_ON(1);
1512 return;
1513 }
1514 }
1515
1516 rb_link_node(&rs->rs_node, parent, newn);
1517 rb_insert_color(&rs->rs_node, &rgd->rd_rstree);
1518
1519 /* Do our rgrp accounting for the reservation */
1520 rgd->rd_requested += rs->rs_requested; /* blocks requested */
1521 spin_unlock(&rgd->rd_rsspin);
1522 trace_gfs2_rs(rs, TRACE_RS_INSERT);
1523 }
1524
1525 /**
1526 * rgd_free - return the number of free blocks we can allocate
1527 * @rgd: the resource group
1528 * @rs: The reservation to free
1529 *
1530 * This function returns the number of free blocks for an rgrp.
1531 * That's the clone-free blocks (blocks that are free, not including those
1532 * still being used for unlinked files that haven't been deleted.)
1533 *
1534 * It also subtracts any blocks reserved by someone else, but does not
1535 * include free blocks that are still part of our current reservation,
1536 * because obviously we can (and will) allocate them.
1537 */
1538 static inline u32 rgd_free(struct gfs2_rgrpd *rgd, struct gfs2_blkreserv *rs)
1539 {
1540 u32 tot_reserved, tot_free;
1541
1542 if (WARN_ON_ONCE(rgd->rd_requested < rs->rs_requested))
1543 return 0;
1544 tot_reserved = rgd->rd_requested - rs->rs_requested;
1545
1546 if (rgd->rd_free_clone < tot_reserved)
1547 tot_reserved = 0;
1548
1549 tot_free = rgd->rd_free_clone - tot_reserved;
1550
1551 return tot_free;
1552 }
1553
1554 /**
1555 * rg_mblk_search - find a group of multiple free blocks to form a reservation
1556 * @rgd: the resource group descriptor
1557 * @ip: pointer to the inode for which we're reserving blocks
1558 * @ap: the allocation parameters
1559 *
1560 */
1561
1562 static void rg_mblk_search(struct gfs2_rgrpd *rgd, struct gfs2_inode *ip,
1563 const struct gfs2_alloc_parms *ap)
1564 {
1565 struct gfs2_rbm rbm = { .rgd = rgd, };
1566 u64 goal;
1567 struct gfs2_blkreserv *rs = &ip->i_res;
1568 u32 extlen;
1569 u32 free_blocks, blocks_available;
1570 int ret;
1571 struct inode *inode = &ip->i_inode;
1572
1573 spin_lock(&rgd->rd_rsspin);
1574 free_blocks = rgd_free(rgd, rs);
1575 if (rgd->rd_free_clone < rgd->rd_requested)
1576 free_blocks = 0;
1577 blocks_available = rgd->rd_free_clone - rgd->rd_reserved;
1578 if (rgd == rs->rs_rgd)
1579 blocks_available += rs->rs_reserved;
1580 spin_unlock(&rgd->rd_rsspin);
1581
1582 if (S_ISDIR(inode->i_mode))
1583 extlen = 1;
1584 else {
1585 extlen = max_t(u32, atomic_read(&ip->i_sizehint), ap->target);
1586 extlen = clamp(extlen, (u32)RGRP_RSRV_MINBLKS, free_blocks);
1587 }
1588 if (free_blocks < extlen || blocks_available < extlen)
1589 return;
1590
1591 /* Find bitmap block that contains bits for goal block */
1592 if (rgrp_contains_block(rgd, ip->i_goal))
1593 goal = ip->i_goal;
1594 else
1595 goal = rgd->rd_last_alloc + rgd->rd_data0;
1596
1597 if (WARN_ON(gfs2_rbm_from_block(&rbm, goal)))
1598 return;
1599
1600 ret = gfs2_rbm_find(&rbm, GFS2_BLKST_FREE, &extlen, &ip->i_res, true);
1601 if (ret == 0) {
1602 rs->rs_start = gfs2_rbm_to_block(&rbm);
1603 rs->rs_requested = extlen;
1604 rs_insert(ip);
1605 } else {
1606 if (goal == rgd->rd_last_alloc + rgd->rd_data0)
1607 rgd->rd_last_alloc = 0;
1608 }
1609 }
1610
1611 /**
1612 * gfs2_next_unreserved_block - Return next block that is not reserved
1613 * @rgd: The resource group
1614 * @block: The starting block
1615 * @length: The required length
1616 * @ignore_rs: Reservation to ignore
1617 *
1618 * If the block does not appear in any reservation, then return the
1619 * block number unchanged. If it does appear in the reservation, then
1620 * keep looking through the tree of reservations in order to find the
1621 * first block number which is not reserved.
1622 */
1623
1624 static u64 gfs2_next_unreserved_block(struct gfs2_rgrpd *rgd, u64 block,
1625 u32 length,
1626 struct gfs2_blkreserv *ignore_rs)
1627 {
1628 struct gfs2_blkreserv *rs;
1629 struct rb_node *n;
1630 int rc;
1631
1632 spin_lock(&rgd->rd_rsspin);
1633 n = rgd->rd_rstree.rb_node;
1634 while (n) {
1635 rs = rb_entry(n, struct gfs2_blkreserv, rs_node);
1636 rc = rs_cmp(block, length, rs);
1637 if (rc < 0)
1638 n = n->rb_left;
1639 else if (rc > 0)
1640 n = n->rb_right;
1641 else
1642 break;
1643 }
1644
1645 if (n) {
1646 while (rs_cmp(block, length, rs) == 0 && rs != ignore_rs) {
1647 block = rs->rs_start + rs->rs_requested;
1648 n = n->rb_right;
1649 if (n == NULL)
1650 break;
1651 rs = rb_entry(n, struct gfs2_blkreserv, rs_node);
1652 }
1653 }
1654
1655 spin_unlock(&rgd->rd_rsspin);
1656 return block;
1657 }
1658
1659 /**
1660 * gfs2_reservation_check_and_update - Check for reservations during block alloc
1661 * @rbm: The current position in the resource group
1662 * @rs: Our own reservation
1663 * @minext: The minimum extent length
1664 * @maxext: A pointer to the maximum extent structure
1665 *
1666 * This checks the current position in the rgrp to see whether there is
1667 * a reservation covering this block. If not then this function is a
1668 * no-op. If there is, then the position is moved to the end of the
1669 * contiguous reservation(s) so that we are pointing at the first
1670 * non-reserved block.
1671 *
1672 * Returns: 0 if no reservation, 1 if @rbm has changed, otherwise an error
1673 */
1674
1675 static int gfs2_reservation_check_and_update(struct gfs2_rbm *rbm,
1676 struct gfs2_blkreserv *rs,
1677 u32 minext,
1678 struct gfs2_extent *maxext)
1679 {
1680 u64 block = gfs2_rbm_to_block(rbm);
1681 u32 extlen = 1;
1682 u64 nblock;
1683
1684 /*
1685 * If we have a minimum extent length, then skip over any extent
1686 * which is less than the min extent length in size.
1687 */
1688 if (minext > 1) {
1689 extlen = gfs2_free_extlen(rbm, minext);
1690 if (extlen <= maxext->len)
1691 goto fail;
1692 }
1693
1694 /*
1695 * Check the extent which has been found against the reservations
1696 * and skip if parts of it are already reserved
1697 */
1698 nblock = gfs2_next_unreserved_block(rbm->rgd, block, extlen, rs);
1699 if (nblock == block) {
1700 if (!minext || extlen >= minext)
1701 return 0;
1702
1703 if (extlen > maxext->len) {
1704 maxext->len = extlen;
1705 maxext->rbm = *rbm;
1706 }
1707 } else {
1708 u64 len = nblock - block;
1709 if (len >= (u64)1 << 32)
1710 return -E2BIG;
1711 extlen = len;
1712 }
1713 fail:
1714 if (gfs2_rbm_add(rbm, extlen))
1715 return -E2BIG;
1716 return 1;
1717 }
1718
1719 /**
1720 * gfs2_rbm_find - Look for blocks of a particular state
1721 * @rbm: Value/result starting position and final position
1722 * @state: The state which we want to find
1723 * @minext: Pointer to the requested extent length
1724 * This is updated to be the actual reservation size.
1725 * @rs: Our own reservation (NULL to skip checking for reservations)
1726 * @nowrap: Stop looking at the end of the rgrp, rather than wrapping
1727 * around until we've reached the starting point.
1728 *
1729 * Side effects:
1730 * - If looking for free blocks, we set GBF_FULL on each bitmap which
1731 * has no free blocks in it.
1732 * - If looking for free blocks, we set rd_extfail_pt on each rgrp which
1733 * has come up short on a free block search.
1734 *
1735 * Returns: 0 on success, -ENOSPC if there is no block of the requested state
1736 */
1737
1738 static int gfs2_rbm_find(struct gfs2_rbm *rbm, u8 state, u32 *minext,
1739 struct gfs2_blkreserv *rs, bool nowrap)
1740 {
1741 bool scan_from_start = rbm->bii == 0 && rbm->offset == 0;
1742 struct buffer_head *bh;
1743 int last_bii;
1744 u32 offset;
1745 u8 *buffer;
1746 bool wrapped = false;
1747 int ret;
1748 struct gfs2_bitmap *bi;
1749 struct gfs2_extent maxext = { .rbm.rgd = rbm->rgd, };
1750
1751 /*
1752 * Determine the last bitmap to search. If we're not starting at the
1753 * beginning of a bitmap, we need to search that bitmap twice to scan
1754 * the entire resource group.
1755 */
1756 last_bii = rbm->bii - (rbm->offset == 0);
1757
1758 while(1) {
1759 bi = rbm_bi(rbm);
1760 if (test_bit(GBF_FULL, &bi->bi_flags) &&
1761 (state == GFS2_BLKST_FREE))
1762 goto next_bitmap;
1763
1764 bh = bi->bi_bh;
1765 buffer = bh->b_data + bi->bi_offset;
1766 WARN_ON(!buffer_uptodate(bh));
1767 if (state != GFS2_BLKST_UNLINKED && bi->bi_clone)
1768 buffer = bi->bi_clone + bi->bi_offset;
1769 offset = gfs2_bitfit(buffer, bi->bi_bytes, rbm->offset, state);
1770 if (offset == BFITNOENT) {
1771 if (state == GFS2_BLKST_FREE && rbm->offset == 0)
1772 set_bit(GBF_FULL, &bi->bi_flags);
1773 goto next_bitmap;
1774 }
1775 rbm->offset = offset;
1776 if (!rs || !minext)
1777 return 0;
1778
1779 ret = gfs2_reservation_check_and_update(rbm, rs, *minext,
1780 &maxext);
1781 if (ret == 0)
1782 return 0;
1783 if (ret > 0)
1784 goto next_iter;
1785 if (ret == -E2BIG) {
1786 rbm->bii = 0;
1787 rbm->offset = 0;
1788 goto res_covered_end_of_rgrp;
1789 }
1790 return ret;
1791
1792 next_bitmap: /* Find next bitmap in the rgrp */
1793 rbm->offset = 0;
1794 rbm->bii++;
1795 if (rbm->bii == rbm->rgd->rd_length)
1796 rbm->bii = 0;
1797 res_covered_end_of_rgrp:
1798 if (rbm->bii == 0) {
1799 if (wrapped)
1800 break;
1801 wrapped = true;
1802 if (nowrap)
1803 break;
1804 }
1805 next_iter:
1806 /* Have we scanned the entire resource group? */
1807 if (wrapped && rbm->bii > last_bii)
1808 break;
1809 }
1810
1811 if (state != GFS2_BLKST_FREE)
1812 return -ENOSPC;
1813
1814 /* If the extent was too small, and it's smaller than the smallest
1815 to have failed before, remember for future reference that it's
1816 useless to search this rgrp again for this amount or more. */
1817 if (wrapped && (scan_from_start || rbm->bii > last_bii) &&
1818 *minext < rbm->rgd->rd_extfail_pt)
1819 rbm->rgd->rd_extfail_pt = *minext - 1;
1820
1821 /* If the maximum extent we found is big enough to fulfill the
1822 minimum requirements, use it anyway. */
1823 if (maxext.len) {
1824 *rbm = maxext.rbm;
1825 *minext = maxext.len;
1826 return 0;
1827 }
1828
1829 return -ENOSPC;
1830 }
1831
1832 /**
1833 * try_rgrp_unlink - Look for any unlinked, allocated, but unused inodes
1834 * @rgd: The rgrp
1835 * @last_unlinked: block address of the last dinode we unlinked
1836 * @skip: block address we should explicitly not unlink
1837 *
1838 * Returns: 0 if no error
1839 * The inode, if one has been found, in inode.
1840 */
1841
1842 static void try_rgrp_unlink(struct gfs2_rgrpd *rgd, u64 *last_unlinked, u64 skip)
1843 {
1844 u64 block;
1845 struct gfs2_sbd *sdp = rgd->rd_sbd;
1846 struct gfs2_glock *gl;
1847 struct gfs2_inode *ip;
1848 int error;
1849 int found = 0;
1850 struct gfs2_rbm rbm = { .rgd = rgd, .bii = 0, .offset = 0 };
1851
1852 while (1) {
1853 error = gfs2_rbm_find(&rbm, GFS2_BLKST_UNLINKED, NULL, NULL,
1854 true);
1855 if (error == -ENOSPC)
1856 break;
1857 if (WARN_ON_ONCE(error))
1858 break;
1859
1860 block = gfs2_rbm_to_block(&rbm);
1861 if (gfs2_rbm_from_block(&rbm, block + 1))
1862 break;
1863 if (*last_unlinked != NO_BLOCK && block <= *last_unlinked)
1864 continue;
1865 if (block == skip)
1866 continue;
1867 *last_unlinked = block;
1868
1869 error = gfs2_glock_get(sdp, block, &gfs2_iopen_glops, CREATE, &gl);
1870 if (error)
1871 continue;
1872
1873 /* If the inode is already in cache, we can ignore it here
1874 * because the existing inode disposal code will deal with
1875 * it when all refs have gone away. Accessing gl_object like
1876 * this is not safe in general. Here it is ok because we do
1877 * not dereference the pointer, and we only need an approx
1878 * answer to whether it is NULL or not.
1879 */
1880 ip = gl->gl_object;
1881
1882 if (ip || !gfs2_queue_try_to_evict(gl))
1883 gfs2_glock_put(gl);
1884 else
1885 found++;
1886
1887 /* Limit reclaim to sensible number of tasks */
1888 if (found > NR_CPUS)
1889 return;
1890 }
1891
1892 rgd->rd_flags &= ~GFS2_RDF_CHECK;
1893 return;
1894 }
1895
1896 /**
1897 * gfs2_rgrp_congested - Use stats to figure out whether an rgrp is congested
1898 * @rgd: The rgrp in question
1899 * @loops: An indication of how picky we can be (0=very, 1=less so)
1900 *
1901 * This function uses the recently added glock statistics in order to
1902 * figure out whether a parciular resource group is suffering from
1903 * contention from multiple nodes. This is done purely on the basis
1904 * of timings, since this is the only data we have to work with and
1905 * our aim here is to reject a resource group which is highly contended
1906 * but (very important) not to do this too often in order to ensure that
1907 * we do not land up introducing fragmentation by changing resource
1908 * groups when not actually required.
1909 *
1910 * The calculation is fairly simple, we want to know whether the SRTTB
1911 * (i.e. smoothed round trip time for blocking operations) to acquire
1912 * the lock for this rgrp's glock is significantly greater than the
1913 * time taken for resource groups on average. We introduce a margin in
1914 * the form of the variable @var which is computed as the sum of the two
1915 * respective variences, and multiplied by a factor depending on @loops
1916 * and whether we have a lot of data to base the decision on. This is
1917 * then tested against the square difference of the means in order to
1918 * decide whether the result is statistically significant or not.
1919 *
1920 * Returns: A boolean verdict on the congestion status
1921 */
1922
1923 static bool gfs2_rgrp_congested(const struct gfs2_rgrpd *rgd, int loops)
1924 {
1925 const struct gfs2_glock *gl = rgd->rd_gl;
1926 const struct gfs2_sbd *sdp = gl->gl_name.ln_sbd;
1927 struct gfs2_lkstats *st;
1928 u64 r_dcount, l_dcount;
1929 u64 l_srttb, a_srttb = 0;
1930 s64 srttb_diff;
1931 u64 sqr_diff;
1932 u64 var;
1933 int cpu, nonzero = 0;
1934
1935 preempt_disable();
1936 for_each_present_cpu(cpu) {
1937 st = &per_cpu_ptr(sdp->sd_lkstats, cpu)->lkstats[LM_TYPE_RGRP];
1938 if (st->stats[GFS2_LKS_SRTTB]) {
1939 a_srttb += st->stats[GFS2_LKS_SRTTB];
1940 nonzero++;
1941 }
1942 }
1943 st = &this_cpu_ptr(sdp->sd_lkstats)->lkstats[LM_TYPE_RGRP];
1944 if (nonzero)
1945 do_div(a_srttb, nonzero);
1946 r_dcount = st->stats[GFS2_LKS_DCOUNT];
1947 var = st->stats[GFS2_LKS_SRTTVARB] +
1948 gl->gl_stats.stats[GFS2_LKS_SRTTVARB];
1949 preempt_enable();
1950
1951 l_srttb = gl->gl_stats.stats[GFS2_LKS_SRTTB];
1952 l_dcount = gl->gl_stats.stats[GFS2_LKS_DCOUNT];
1953
1954 if ((l_dcount < 1) || (r_dcount < 1) || (a_srttb == 0))
1955 return false;
1956
1957 srttb_diff = a_srttb - l_srttb;
1958 sqr_diff = srttb_diff * srttb_diff;
1959
1960 var *= 2;
1961 if (l_dcount < 8 || r_dcount < 8)
1962 var *= 2;
1963 if (loops == 1)
1964 var *= 2;
1965
1966 return ((srttb_diff < 0) && (sqr_diff > var));
1967 }
1968
1969 /**
1970 * gfs2_rgrp_used_recently - test if an rgrp has been used recently
1971 * @rs: The block reservation with the rgrp to test
1972 * @msecs: The time limit in milliseconds
1973 *
1974 * Returns: True if the rgrp glock has been used within the time limit
1975 */
1976 static bool gfs2_rgrp_used_recently(const struct gfs2_blkreserv *rs,
1977 u64 msecs)
1978 {
1979 u64 tdiff;
1980
1981 tdiff = ktime_to_ns(ktime_sub(ktime_get_real(),
1982 rs->rs_rgd->rd_gl->gl_dstamp));
1983
1984 return tdiff > (msecs * 1000 * 1000);
1985 }
1986
1987 static u32 gfs2_orlov_skip(const struct gfs2_inode *ip)
1988 {
1989 const struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
1990 u32 skip;
1991
1992 get_random_bytes(&skip, sizeof(skip));
1993 return skip % sdp->sd_rgrps;
1994 }
1995
1996 static bool gfs2_select_rgrp(struct gfs2_rgrpd **pos, const struct gfs2_rgrpd *begin)
1997 {
1998 struct gfs2_rgrpd *rgd = *pos;
1999 struct gfs2_sbd *sdp = rgd->rd_sbd;
2000
2001 rgd = gfs2_rgrpd_get_next(rgd);
2002 if (rgd == NULL)
2003 rgd = gfs2_rgrpd_get_first(sdp);
2004 *pos = rgd;
2005 if (rgd != begin) /* If we didn't wrap */
2006 return true;
2007 return false;
2008 }
2009
2010 /**
2011 * fast_to_acquire - determine if a resource group will be fast to acquire
2012 * @rgd: The rgrp
2013 *
2014 * If this is one of our preferred rgrps, it should be quicker to acquire,
2015 * because we tried to set ourselves up as dlm lock master.
2016 */
2017 static inline int fast_to_acquire(struct gfs2_rgrpd *rgd)
2018 {
2019 struct gfs2_glock *gl = rgd->rd_gl;
2020
2021 if (gl->gl_state != LM_ST_UNLOCKED && list_empty(&gl->gl_holders) &&
2022 !test_bit(GLF_DEMOTE_IN_PROGRESS, &gl->gl_flags) &&
2023 !test_bit(GLF_DEMOTE, &gl->gl_flags))
2024 return 1;
2025 if (rgd->rd_flags & GFS2_RDF_PREFERRED)
2026 return 1;
2027 return 0;
2028 }
2029
2030 /**
2031 * gfs2_inplace_reserve - Reserve space in the filesystem
2032 * @ip: the inode to reserve space for
2033 * @ap: the allocation parameters
2034 *
2035 * We try our best to find an rgrp that has at least ap->target blocks
2036 * available. After a couple of passes (loops == 2), the prospects of finding
2037 * such an rgrp diminish. At this stage, we return the first rgrp that has
2038 * at least ap->min_target blocks available.
2039 *
2040 * Returns: 0 on success,
2041 * -ENOMEM if a suitable rgrp can't be found
2042 * errno otherwise
2043 */
2044
2045 int gfs2_inplace_reserve(struct gfs2_inode *ip, struct gfs2_alloc_parms *ap)
2046 {
2047 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
2048 struct gfs2_rgrpd *begin = NULL;
2049 struct gfs2_blkreserv *rs = &ip->i_res;
2050 int error = 0, flags = LM_FLAG_NODE_SCOPE;
2051 bool rg_locked;
2052 u64 last_unlinked = NO_BLOCK;
2053 u32 target = ap->target;
2054 int loops = 0;
2055 u32 free_blocks, blocks_available, skip = 0;
2056
2057 BUG_ON(rs->rs_reserved);
2058
2059 if (sdp->sd_args.ar_rgrplvb)
2060 flags |= GL_SKIP;
2061 if (gfs2_assert_warn(sdp, target))
2062 return -EINVAL;
2063 if (gfs2_rs_active(rs)) {
2064 begin = rs->rs_rgd;
2065 } else if (rs->rs_rgd &&
2066 rgrp_contains_block(rs->rs_rgd, ip->i_goal)) {
2067 begin = rs->rs_rgd;
2068 } else {
2069 check_and_update_goal(ip);
2070 rs->rs_rgd = begin = gfs2_blk2rgrpd(sdp, ip->i_goal, 1);
2071 }
2072 if (S_ISDIR(ip->i_inode.i_mode) && (ap->aflags & GFS2_AF_ORLOV))
2073 skip = gfs2_orlov_skip(ip);
2074 if (rs->rs_rgd == NULL)
2075 return -EBADSLT;
2076
2077 while (loops < 3) {
2078 struct gfs2_rgrpd *rgd;
2079
2080 rg_locked = gfs2_glock_is_locked_by_me(rs->rs_rgd->rd_gl);
2081 if (rg_locked) {
2082 rgrp_lock_local(rs->rs_rgd);
2083 } else {
2084 if (skip && skip--)
2085 goto next_rgrp;
2086 if (!gfs2_rs_active(rs)) {
2087 if (loops == 0 &&
2088 !fast_to_acquire(rs->rs_rgd))
2089 goto next_rgrp;
2090 if ((loops < 2) &&
2091 gfs2_rgrp_used_recently(rs, 1000) &&
2092 gfs2_rgrp_congested(rs->rs_rgd, loops))
2093 goto next_rgrp;
2094 }
2095 error = gfs2_glock_nq_init(rs->rs_rgd->rd_gl,
2096 LM_ST_EXCLUSIVE, flags,
2097 &ip->i_rgd_gh);
2098 if (unlikely(error))
2099 return error;
2100 rgrp_lock_local(rs->rs_rgd);
2101 if (!gfs2_rs_active(rs) && (loops < 2) &&
2102 gfs2_rgrp_congested(rs->rs_rgd, loops))
2103 goto skip_rgrp;
2104 if (sdp->sd_args.ar_rgrplvb) {
2105 error = update_rgrp_lvb(rs->rs_rgd,
2106 &ip->i_rgd_gh);
2107 if (unlikely(error)) {
2108 rgrp_unlock_local(rs->rs_rgd);
2109 gfs2_glock_dq_uninit(&ip->i_rgd_gh);
2110 return error;
2111 }
2112 }
2113 }
2114
2115 /* Skip unusable resource groups */
2116 if ((rs->rs_rgd->rd_flags & (GFS2_RGF_NOALLOC |
2117 GFS2_RDF_ERROR)) ||
2118 (loops == 0 && target > rs->rs_rgd->rd_extfail_pt))
2119 goto skip_rgrp;
2120
2121 if (sdp->sd_args.ar_rgrplvb) {
2122 error = gfs2_instantiate(&ip->i_rgd_gh);
2123 if (error)
2124 goto skip_rgrp;
2125 }
2126
2127 /* Get a reservation if we don't already have one */
2128 if (!gfs2_rs_active(rs))
2129 rg_mblk_search(rs->rs_rgd, ip, ap);
2130
2131 /* Skip rgrps when we can't get a reservation on first pass */
2132 if (!gfs2_rs_active(rs) && (loops < 1))
2133 goto check_rgrp;
2134
2135 /* If rgrp has enough free space, use it */
2136 rgd = rs->rs_rgd;
2137 spin_lock(&rgd->rd_rsspin);
2138 free_blocks = rgd_free(rgd, rs);
2139 blocks_available = rgd->rd_free_clone - rgd->rd_reserved;
2140 if (free_blocks < target || blocks_available < target) {
2141 spin_unlock(&rgd->rd_rsspin);
2142 goto check_rgrp;
2143 }
2144 rs->rs_reserved = ap->target;
2145 if (rs->rs_reserved > blocks_available)
2146 rs->rs_reserved = blocks_available;
2147 rgd->rd_reserved += rs->rs_reserved;
2148 spin_unlock(&rgd->rd_rsspin);
2149 rgrp_unlock_local(rs->rs_rgd);
2150 return 0;
2151 check_rgrp:
2152 /* Check for unlinked inodes which can be reclaimed */
2153 if (rs->rs_rgd->rd_flags & GFS2_RDF_CHECK)
2154 try_rgrp_unlink(rs->rs_rgd, &last_unlinked,
2155 ip->i_no_addr);
2156 skip_rgrp:
2157 rgrp_unlock_local(rs->rs_rgd);
2158
2159 /* Drop reservation, if we couldn't use reserved rgrp */
2160 if (gfs2_rs_active(rs))
2161 gfs2_rs_deltree(rs);
2162
2163 /* Unlock rgrp if required */
2164 if (!rg_locked)
2165 gfs2_glock_dq_uninit(&ip->i_rgd_gh);
2166 next_rgrp:
2167 /* Find the next rgrp, and continue looking */
2168 if (gfs2_select_rgrp(&rs->rs_rgd, begin))
2169 continue;
2170 if (skip)
2171 continue;
2172
2173 /* If we've scanned all the rgrps, but found no free blocks
2174 * then this checks for some less likely conditions before
2175 * trying again.
2176 */
2177 loops++;
2178 /* Check that fs hasn't grown if writing to rindex */
2179 if (ip == GFS2_I(sdp->sd_rindex) && !sdp->sd_rindex_uptodate) {
2180 error = gfs2_ri_update(ip);
2181 if (error)
2182 return error;
2183 }
2184 /* Flushing the log may release space */
2185 if (loops == 2) {
2186 if (ap->min_target)
2187 target = ap->min_target;
2188 gfs2_log_flush(sdp, NULL, GFS2_LOG_HEAD_FLUSH_NORMAL |
2189 GFS2_LFC_INPLACE_RESERVE);
2190 }
2191 }
2192
2193 return -ENOSPC;
2194 }
2195
2196 /**
2197 * gfs2_inplace_release - release an inplace reservation
2198 * @ip: the inode the reservation was taken out on
2199 *
2200 * Release a reservation made by gfs2_inplace_reserve().
2201 */
2202
2203 void gfs2_inplace_release(struct gfs2_inode *ip)
2204 {
2205 struct gfs2_blkreserv *rs = &ip->i_res;
2206
2207 if (rs->rs_reserved) {
2208 struct gfs2_rgrpd *rgd = rs->rs_rgd;
2209
2210 spin_lock(&rgd->rd_rsspin);
2211 GLOCK_BUG_ON(rgd->rd_gl, rgd->rd_reserved < rs->rs_reserved);
2212 rgd->rd_reserved -= rs->rs_reserved;
2213 spin_unlock(&rgd->rd_rsspin);
2214 rs->rs_reserved = 0;
2215 }
2216 if (gfs2_holder_initialized(&ip->i_rgd_gh))
2217 gfs2_glock_dq_uninit(&ip->i_rgd_gh);
2218 }
2219
2220 /**
2221 * gfs2_alloc_extent - allocate an extent from a given bitmap
2222 * @rbm: the resource group information
2223 * @dinode: TRUE if the first block we allocate is for a dinode
2224 * @n: The extent length (value/result)
2225 *
2226 * Add the bitmap buffer to the transaction.
2227 * Set the found bits to @new_state to change block's allocation state.
2228 */
2229 static void gfs2_alloc_extent(const struct gfs2_rbm *rbm, bool dinode,
2230 unsigned int *n)
2231 {
2232 struct gfs2_rbm pos = { .rgd = rbm->rgd, };
2233 const unsigned int elen = *n;
2234 u64 block;
2235 int ret;
2236
2237 *n = 1;
2238 block = gfs2_rbm_to_block(rbm);
2239 gfs2_trans_add_meta(rbm->rgd->rd_gl, rbm_bi(rbm)->bi_bh);
2240 gfs2_setbit(rbm, true, dinode ? GFS2_BLKST_DINODE : GFS2_BLKST_USED);
2241 block++;
2242 while (*n < elen) {
2243 ret = gfs2_rbm_from_block(&pos, block);
2244 if (ret || gfs2_testbit(&pos, true) != GFS2_BLKST_FREE)
2245 break;
2246 gfs2_trans_add_meta(pos.rgd->rd_gl, rbm_bi(&pos)->bi_bh);
2247 gfs2_setbit(&pos, true, GFS2_BLKST_USED);
2248 (*n)++;
2249 block++;
2250 }
2251 }
2252
2253 /**
2254 * rgblk_free - Change alloc state of given block(s)
2255 * @sdp: the filesystem
2256 * @rgd: the resource group the blocks are in
2257 * @bstart: the start of a run of blocks to free
2258 * @blen: the length of the block run (all must lie within ONE RG!)
2259 * @new_state: GFS2_BLKST_XXX the after-allocation block state
2260 */
2261
2262 static void rgblk_free(struct gfs2_sbd *sdp, struct gfs2_rgrpd *rgd,
2263 u64 bstart, u32 blen, unsigned char new_state)
2264 {
2265 struct gfs2_rbm rbm;
2266 struct gfs2_bitmap *bi, *bi_prev = NULL;
2267
2268 rbm.rgd = rgd;
2269 if (WARN_ON_ONCE(gfs2_rbm_from_block(&rbm, bstart)))
2270 return;
2271 while (blen--) {
2272 bi = rbm_bi(&rbm);
2273 if (bi != bi_prev) {
2274 if (!bi->bi_clone) {
2275 bi->bi_clone = kmalloc(bi->bi_bh->b_size,
2276 GFP_NOFS | __GFP_NOFAIL);
2277 memcpy(bi->bi_clone + bi->bi_offset,
2278 bi->bi_bh->b_data + bi->bi_offset,
2279 bi->bi_bytes);
2280 }
2281 gfs2_trans_add_meta(rbm.rgd->rd_gl, bi->bi_bh);
2282 bi_prev = bi;
2283 }
2284 gfs2_setbit(&rbm, false, new_state);
2285 gfs2_rbm_add(&rbm, 1);
2286 }
2287 }
2288
2289 /**
2290 * gfs2_rgrp_dump - print out an rgrp
2291 * @seq: The iterator
2292 * @rgd: The rgrp in question
2293 * @fs_id_buf: pointer to file system id (if requested)
2294 *
2295 */
2296
2297 void gfs2_rgrp_dump(struct seq_file *seq, struct gfs2_rgrpd *rgd,
2298 const char *fs_id_buf)
2299 {
2300 struct gfs2_blkreserv *trs;
2301 const struct rb_node *n;
2302
2303 spin_lock(&rgd->rd_rsspin);
2304 gfs2_print_dbg(seq, "%s R: n:%llu f:%02x b:%u/%u i:%u q:%u r:%u e:%u\n",
2305 fs_id_buf,
2306 (unsigned long long)rgd->rd_addr, rgd->rd_flags,
2307 rgd->rd_free, rgd->rd_free_clone, rgd->rd_dinodes,
2308 rgd->rd_requested, rgd->rd_reserved, rgd->rd_extfail_pt);
2309 if (rgd->rd_sbd->sd_args.ar_rgrplvb && rgd->rd_rgl) {
2310 struct gfs2_rgrp_lvb *rgl = rgd->rd_rgl;
2311
2312 gfs2_print_dbg(seq, "%s L: f:%02x b:%u i:%u\n", fs_id_buf,
2313 be32_to_cpu(rgl->rl_flags),
2314 be32_to_cpu(rgl->rl_free),
2315 be32_to_cpu(rgl->rl_dinodes));
2316 }
2317 for (n = rb_first(&rgd->rd_rstree); n; n = rb_next(&trs->rs_node)) {
2318 trs = rb_entry(n, struct gfs2_blkreserv, rs_node);
2319 dump_rs(seq, trs, fs_id_buf);
2320 }
2321 spin_unlock(&rgd->rd_rsspin);
2322 }
2323
2324 static void gfs2_rgrp_error(struct gfs2_rgrpd *rgd)
2325 {
2326 struct gfs2_sbd *sdp = rgd->rd_sbd;
2327 char fs_id_buf[sizeof(sdp->sd_fsname) + 7];
2328
2329 fs_warn(sdp, "rgrp %llu has an error, marking it readonly until umount\n",
2330 (unsigned long long)rgd->rd_addr);
2331 fs_warn(sdp, "umount on all nodes and run fsck.gfs2 to fix the error\n");
2332 sprintf(fs_id_buf, "fsid=%s: ", sdp->sd_fsname);
2333 gfs2_rgrp_dump(NULL, rgd, fs_id_buf);
2334 rgd->rd_flags |= GFS2_RDF_ERROR;
2335 }
2336
2337 /**
2338 * gfs2_adjust_reservation - Adjust (or remove) a reservation after allocation
2339 * @ip: The inode we have just allocated blocks for
2340 * @rbm: The start of the allocated blocks
2341 * @len: The extent length
2342 *
2343 * Adjusts a reservation after an allocation has taken place. If the
2344 * reservation does not match the allocation, or if it is now empty
2345 * then it is removed.
2346 */
2347
2348 static void gfs2_adjust_reservation(struct gfs2_inode *ip,
2349 const struct gfs2_rbm *rbm, unsigned len)
2350 {
2351 struct gfs2_blkreserv *rs = &ip->i_res;
2352 struct gfs2_rgrpd *rgd = rbm->rgd;
2353
2354 BUG_ON(rs->rs_reserved < len);
2355 rs->rs_reserved -= len;
2356 if (gfs2_rs_active(rs)) {
2357 u64 start = gfs2_rbm_to_block(rbm);
2358
2359 if (rs->rs_start == start) {
2360 unsigned int rlen;
2361
2362 rs->rs_start += len;
2363 rlen = min(rs->rs_requested, len);
2364 rs->rs_requested -= rlen;
2365 rgd->rd_requested -= rlen;
2366 trace_gfs2_rs(rs, TRACE_RS_CLAIM);
2367 if (rs->rs_start < rgd->rd_data0 + rgd->rd_data &&
2368 rs->rs_requested)
2369 return;
2370 /* We used up our block reservation, so we should
2371 reserve more blocks next time. */
2372 atomic_add(RGRP_RSRV_ADDBLKS, &ip->i_sizehint);
2373 }
2374 __rs_deltree(rs);
2375 }
2376 }
2377
2378 /**
2379 * gfs2_set_alloc_start - Set starting point for block allocation
2380 * @rbm: The rbm which will be set to the required location
2381 * @ip: The gfs2 inode
2382 * @dinode: Flag to say if allocation includes a new inode
2383 *
2384 * This sets the starting point from the reservation if one is active
2385 * otherwise it falls back to guessing a start point based on the
2386 * inode's goal block or the last allocation point in the rgrp.
2387 */
2388
2389 static void gfs2_set_alloc_start(struct gfs2_rbm *rbm,
2390 const struct gfs2_inode *ip, bool dinode)
2391 {
2392 u64 goal;
2393
2394 if (gfs2_rs_active(&ip->i_res)) {
2395 goal = ip->i_res.rs_start;
2396 } else {
2397 if (!dinode && rgrp_contains_block(rbm->rgd, ip->i_goal))
2398 goal = ip->i_goal;
2399 else
2400 goal = rbm->rgd->rd_last_alloc + rbm->rgd->rd_data0;
2401 }
2402 if (WARN_ON_ONCE(gfs2_rbm_from_block(rbm, goal))) {
2403 rbm->bii = 0;
2404 rbm->offset = 0;
2405 }
2406 }
2407
2408 /**
2409 * gfs2_alloc_blocks - Allocate one or more blocks of data and/or a dinode
2410 * @ip: the inode to allocate the block for
2411 * @bn: Used to return the starting block number
2412 * @nblocks: requested number of blocks/extent length (value/result)
2413 * @dinode: 1 if we're allocating a dinode block, else 0
2414 *
2415 * Returns: 0 or error
2416 */
2417
2418 int gfs2_alloc_blocks(struct gfs2_inode *ip, u64 *bn, unsigned int *nblocks,
2419 bool dinode)
2420 {
2421 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
2422 struct buffer_head *dibh;
2423 struct gfs2_rbm rbm = { .rgd = ip->i_res.rs_rgd, };
2424 u64 block; /* block, within the file system scope */
2425 u32 minext = 1;
2426 int error = -ENOSPC;
2427
2428 BUG_ON(ip->i_res.rs_reserved < *nblocks);
2429
2430 rgrp_lock_local(rbm.rgd);
2431 if (gfs2_rs_active(&ip->i_res)) {
2432 gfs2_set_alloc_start(&rbm, ip, dinode);
2433 error = gfs2_rbm_find(&rbm, GFS2_BLKST_FREE, &minext, &ip->i_res, false);
2434 }
2435 if (error == -ENOSPC) {
2436 gfs2_set_alloc_start(&rbm, ip, dinode);
2437 error = gfs2_rbm_find(&rbm, GFS2_BLKST_FREE, &minext, NULL, false);
2438 }
2439
2440 /* Since all blocks are reserved in advance, this shouldn't happen */
2441 if (error) {
2442 fs_warn(sdp, "inum=%llu error=%d, nblocks=%u, full=%d fail_pt=%d\n",
2443 (unsigned long long)ip->i_no_addr, error, *nblocks,
2444 test_bit(GBF_FULL, &rbm.rgd->rd_bits->bi_flags),
2445 rbm.rgd->rd_extfail_pt);
2446 goto rgrp_error;
2447 }
2448
2449 gfs2_alloc_extent(&rbm, dinode, nblocks);
2450 block = gfs2_rbm_to_block(&rbm);
2451 rbm.rgd->rd_last_alloc = block - rbm.rgd->rd_data0;
2452 if (!dinode) {
2453 ip->i_goal = block + *nblocks - 1;
2454 error = gfs2_meta_inode_buffer(ip, &dibh);
2455 if (error == 0) {
2456 struct gfs2_dinode *di =
2457 (struct gfs2_dinode *)dibh->b_data;
2458 gfs2_trans_add_meta(ip->i_gl, dibh);
2459 di->di_goal_meta = di->di_goal_data =
2460 cpu_to_be64(ip->i_goal);
2461 brelse(dibh);
2462 }
2463 }
2464 spin_lock(&rbm.rgd->rd_rsspin);
2465 gfs2_adjust_reservation(ip, &rbm, *nblocks);
2466 if (rbm.rgd->rd_free < *nblocks || rbm.rgd->rd_reserved < *nblocks) {
2467 fs_warn(sdp, "nblocks=%u\n", *nblocks);
2468 spin_unlock(&rbm.rgd->rd_rsspin);
2469 goto rgrp_error;
2470 }
2471 GLOCK_BUG_ON(rbm.rgd->rd_gl, rbm.rgd->rd_reserved < *nblocks);
2472 GLOCK_BUG_ON(rbm.rgd->rd_gl, rbm.rgd->rd_free_clone < *nblocks);
2473 GLOCK_BUG_ON(rbm.rgd->rd_gl, rbm.rgd->rd_free < *nblocks);
2474 rbm.rgd->rd_reserved -= *nblocks;
2475 rbm.rgd->rd_free_clone -= *nblocks;
2476 rbm.rgd->rd_free -= *nblocks;
2477 spin_unlock(&rbm.rgd->rd_rsspin);
2478 if (dinode) {
2479 u64 generation;
2480
2481 rbm.rgd->rd_dinodes++;
2482 generation = rbm.rgd->rd_igeneration++;
2483 if (generation == 0)
2484 generation = rbm.rgd->rd_igeneration++;
2485 ip->i_generation = generation;
2486 }
2487
2488 gfs2_trans_add_meta(rbm.rgd->rd_gl, rbm.rgd->rd_bits[0].bi_bh);
2489 gfs2_rgrp_out(rbm.rgd, rbm.rgd->rd_bits[0].bi_bh->b_data);
2490 rgrp_unlock_local(rbm.rgd);
2491
2492 gfs2_statfs_change(sdp, 0, -(s64)*nblocks, dinode ? 1 : 0);
2493 if (dinode)
2494 gfs2_trans_remove_revoke(sdp, block, *nblocks);
2495
2496 gfs2_quota_change(ip, *nblocks, ip->i_inode.i_uid, ip->i_inode.i_gid);
2497
2498 trace_gfs2_block_alloc(ip, rbm.rgd, block, *nblocks,
2499 dinode ? GFS2_BLKST_DINODE : GFS2_BLKST_USED);
2500 *bn = block;
2501 return 0;
2502
2503 rgrp_error:
2504 rgrp_unlock_local(rbm.rgd);
2505 gfs2_rgrp_error(rbm.rgd);
2506 return -EIO;
2507 }
2508
2509 /**
2510 * __gfs2_free_blocks - free a contiguous run of block(s)
2511 * @ip: the inode these blocks are being freed from
2512 * @rgd: the resource group the blocks are in
2513 * @bstart: first block of a run of contiguous blocks
2514 * @blen: the length of the block run
2515 * @meta: 1 if the blocks represent metadata
2516 *
2517 */
2518
2519 void __gfs2_free_blocks(struct gfs2_inode *ip, struct gfs2_rgrpd *rgd,
2520 u64 bstart, u32 blen, int meta)
2521 {
2522 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
2523
2524 rgrp_lock_local(rgd);
2525 rgblk_free(sdp, rgd, bstart, blen, GFS2_BLKST_FREE);
2526 trace_gfs2_block_alloc(ip, rgd, bstart, blen, GFS2_BLKST_FREE);
2527 rgd->rd_free += blen;
2528 rgd->rd_flags &= ~GFS2_RGF_TRIMMED;
2529 gfs2_trans_add_meta(rgd->rd_gl, rgd->rd_bits[0].bi_bh);
2530 gfs2_rgrp_out(rgd, rgd->rd_bits[0].bi_bh->b_data);
2531 rgrp_unlock_local(rgd);
2532
2533 /* Directories keep their data in the metadata address space */
2534 if (meta || ip->i_depth || gfs2_is_jdata(ip))
2535 gfs2_journal_wipe(ip, bstart, blen);
2536 }
2537
2538 /**
2539 * gfs2_free_meta - free a contiguous run of data block(s)
2540 * @ip: the inode these blocks are being freed from
2541 * @rgd: the resource group the blocks are in
2542 * @bstart: first block of a run of contiguous blocks
2543 * @blen: the length of the block run
2544 *
2545 */
2546
2547 void gfs2_free_meta(struct gfs2_inode *ip, struct gfs2_rgrpd *rgd,
2548 u64 bstart, u32 blen)
2549 {
2550 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
2551
2552 __gfs2_free_blocks(ip, rgd, bstart, blen, 1);
2553 gfs2_statfs_change(sdp, 0, +blen, 0);
2554 gfs2_quota_change(ip, -(s64)blen, ip->i_inode.i_uid, ip->i_inode.i_gid);
2555 }
2556
2557 void gfs2_unlink_di(struct inode *inode)
2558 {
2559 struct gfs2_inode *ip = GFS2_I(inode);
2560 struct gfs2_sbd *sdp = GFS2_SB(inode);
2561 struct gfs2_rgrpd *rgd;
2562 u64 blkno = ip->i_no_addr;
2563
2564 rgd = gfs2_blk2rgrpd(sdp, blkno, true);
2565 if (!rgd)
2566 return;
2567 rgrp_lock_local(rgd);
2568 rgblk_free(sdp, rgd, blkno, 1, GFS2_BLKST_UNLINKED);
2569 trace_gfs2_block_alloc(ip, rgd, blkno, 1, GFS2_BLKST_UNLINKED);
2570 gfs2_trans_add_meta(rgd->rd_gl, rgd->rd_bits[0].bi_bh);
2571 gfs2_rgrp_out(rgd, rgd->rd_bits[0].bi_bh->b_data);
2572 be32_add_cpu(&rgd->rd_rgl->rl_unlinked, 1);
2573 rgrp_unlock_local(rgd);
2574 }
2575
2576 void gfs2_free_di(struct gfs2_rgrpd *rgd, struct gfs2_inode *ip)
2577 {
2578 struct gfs2_sbd *sdp = rgd->rd_sbd;
2579
2580 rgrp_lock_local(rgd);
2581 rgblk_free(sdp, rgd, ip->i_no_addr, 1, GFS2_BLKST_FREE);
2582 if (!rgd->rd_dinodes)
2583 gfs2_consist_rgrpd(rgd);
2584 rgd->rd_dinodes--;
2585 rgd->rd_free++;
2586
2587 gfs2_trans_add_meta(rgd->rd_gl, rgd->rd_bits[0].bi_bh);
2588 gfs2_rgrp_out(rgd, rgd->rd_bits[0].bi_bh->b_data);
2589 be32_add_cpu(&rgd->rd_rgl->rl_unlinked, -1);
2590 rgrp_unlock_local(rgd);
2591
2592 gfs2_statfs_change(sdp, 0, +1, -1);
2593 trace_gfs2_block_alloc(ip, rgd, ip->i_no_addr, 1, GFS2_BLKST_FREE);
2594 gfs2_quota_change(ip, -1, ip->i_inode.i_uid, ip->i_inode.i_gid);
2595 gfs2_journal_wipe(ip, ip->i_no_addr, 1);
2596 }
2597
2598 /**
2599 * gfs2_check_blk_type - Check the type of a block
2600 * @sdp: The superblock
2601 * @no_addr: The block number to check
2602 * @type: The block type we are looking for
2603 *
2604 * The inode glock of @no_addr must be held. The @type to check for is either
2605 * GFS2_BLKST_DINODE or GFS2_BLKST_UNLINKED; checking for type GFS2_BLKST_FREE
2606 * or GFS2_BLKST_USED would make no sense.
2607 *
2608 * Returns: 0 if the block type matches the expected type
2609 * -ESTALE if it doesn't match
2610 * or -ve errno if something went wrong while checking
2611 */
2612
2613 int gfs2_check_blk_type(struct gfs2_sbd *sdp, u64 no_addr, unsigned int type)
2614 {
2615 struct gfs2_rgrpd *rgd;
2616 struct gfs2_holder rgd_gh;
2617 struct gfs2_rbm rbm;
2618 int error = -EINVAL;
2619
2620 rgd = gfs2_blk2rgrpd(sdp, no_addr, 1);
2621 if (!rgd)
2622 goto fail;
2623
2624 error = gfs2_glock_nq_init(rgd->rd_gl, LM_ST_SHARED, 0, &rgd_gh);
2625 if (error)
2626 goto fail;
2627
2628 rbm.rgd = rgd;
2629 error = gfs2_rbm_from_block(&rbm, no_addr);
2630 if (!WARN_ON_ONCE(error)) {
2631 /*
2632 * No need to take the local resource group lock here; the
2633 * inode glock of @no_addr provides the necessary
2634 * synchronization in case the block is an inode. (In case
2635 * the block is not an inode, the block type will not match
2636 * the @type we are looking for.)
2637 */
2638 if (gfs2_testbit(&rbm, false) != type)
2639 error = -ESTALE;
2640 }
2641
2642 gfs2_glock_dq_uninit(&rgd_gh);
2643
2644 fail:
2645 return error;
2646 }
2647
2648 /**
2649 * gfs2_rlist_add - add a RG to a list of RGs
2650 * @ip: the inode
2651 * @rlist: the list of resource groups
2652 * @block: the block
2653 *
2654 * Figure out what RG a block belongs to and add that RG to the list
2655 *
2656 * FIXME: Don't use NOFAIL
2657 *
2658 */
2659
2660 void gfs2_rlist_add(struct gfs2_inode *ip, struct gfs2_rgrp_list *rlist,
2661 u64 block)
2662 {
2663 struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
2664 struct gfs2_rgrpd *rgd;
2665 struct gfs2_rgrpd **tmp;
2666 unsigned int new_space;
2667 unsigned int x;
2668
2669 if (gfs2_assert_warn(sdp, !rlist->rl_ghs))
2670 return;
2671
2672 /*
2673 * The resource group last accessed is kept in the last position.
2674 */
2675
2676 if (rlist->rl_rgrps) {
2677 rgd = rlist->rl_rgd[rlist->rl_rgrps - 1];
2678 if (rgrp_contains_block(rgd, block))
2679 return;
2680 rgd = gfs2_blk2rgrpd(sdp, block, 1);
2681 } else {
2682 rgd = ip->i_res.rs_rgd;
2683 if (!rgd || !rgrp_contains_block(rgd, block))
2684 rgd = gfs2_blk2rgrpd(sdp, block, 1);
2685 }
2686
2687 if (!rgd) {
2688 fs_err(sdp, "rlist_add: no rgrp for block %llu\n",
2689 (unsigned long long)block);
2690 return;
2691 }
2692
2693 for (x = 0; x < rlist->rl_rgrps; x++) {
2694 if (rlist->rl_rgd[x] == rgd) {
2695 swap(rlist->rl_rgd[x],
2696 rlist->rl_rgd[rlist->rl_rgrps - 1]);
2697 return;
2698 }
2699 }
2700
2701 if (rlist->rl_rgrps == rlist->rl_space) {
2702 new_space = rlist->rl_space + 10;
2703
2704 tmp = kcalloc(new_space, sizeof(struct gfs2_rgrpd *),
2705 GFP_NOFS | __GFP_NOFAIL);
2706
2707 if (rlist->rl_rgd) {
2708 memcpy(tmp, rlist->rl_rgd,
2709 rlist->rl_space * sizeof(struct gfs2_rgrpd *));
2710 kfree(rlist->rl_rgd);
2711 }
2712
2713 rlist->rl_space = new_space;
2714 rlist->rl_rgd = tmp;
2715 }
2716
2717 rlist->rl_rgd[rlist->rl_rgrps++] = rgd;
2718 }
2719
2720 /**
2721 * gfs2_rlist_alloc - all RGs have been added to the rlist, now allocate
2722 * and initialize an array of glock holders for them
2723 * @rlist: the list of resource groups
2724 * @state: the state we're requesting
2725 * @flags: the modifier flags
2726 *
2727 * FIXME: Don't use NOFAIL
2728 *
2729 */
2730
2731 void gfs2_rlist_alloc(struct gfs2_rgrp_list *rlist,
2732 unsigned int state, u16 flags)
2733 {
2734 unsigned int x;
2735
2736 rlist->rl_ghs = kmalloc_array(rlist->rl_rgrps,
2737 sizeof(struct gfs2_holder),
2738 GFP_NOFS | __GFP_NOFAIL);
2739 for (x = 0; x < rlist->rl_rgrps; x++)
2740 gfs2_holder_init(rlist->rl_rgd[x]->rd_gl, state, flags,
2741 &rlist->rl_ghs[x]);
2742 }
2743
2744 /**
2745 * gfs2_rlist_free - free a resource group list
2746 * @rlist: the list of resource groups
2747 *
2748 */
2749
2750 void gfs2_rlist_free(struct gfs2_rgrp_list *rlist)
2751 {
2752 unsigned int x;
2753
2754 kfree(rlist->rl_rgd);
2755
2756 if (rlist->rl_ghs) {
2757 for (x = 0; x < rlist->rl_rgrps; x++)
2758 gfs2_holder_uninit(&rlist->rl_ghs[x]);
2759 kfree(rlist->rl_ghs);
2760 rlist->rl_ghs = NULL;
2761 }
2762 }
2763
2764 void rgrp_lock_local(struct gfs2_rgrpd *rgd)
2765 {
2766 mutex_lock(&rgd->rd_mutex);
2767 }
2768
2769 void rgrp_unlock_local(struct gfs2_rgrpd *rgd)
2770 {
2771 mutex_unlock(&rgd->rd_mutex);
2772 }
2773
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.