cpu/asm.h: Use optimal alignment for assembly functions on x86_64.
[dragonfly.git] / sys / vfs / hammer2 / hammer2_bulkfree.c
blob07ba7208dbb3f2934f03d4bf59486f9cfcb01c9c
1 /*
2 * Copyright (c) 2013-2015 The DragonFly Project. All rights reserved.
4 * This code is derived from software contributed to The DragonFly Project
5 * by Matthew Dillon <dillon@dragonflybsd.org>
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in
15 * the documentation and/or other materials provided with the
16 * distribution.
17 * 3. Neither the name of The DragonFly Project nor the names of its
18 * contributors may be used to endorse or promote products derived
19 * from this software without specific, prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
24 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
25 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
26 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
27 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
28 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
29 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
30 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
31 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32 * SUCH DAMAGE.
34 #include <sys/param.h>
35 #include <sys/systm.h>
36 #include <sys/kernel.h>
37 #include <sys/fcntl.h>
38 #include <sys/buf.h>
39 #include <sys/proc.h>
40 #include <sys/namei.h>
41 #include <sys/mount.h>
42 #include <sys/vnode.h>
43 #include <sys/mountctl.h>
44 #include <vm/vm_kern.h>
45 #include <vm/vm_extern.h>
47 #include "hammer2.h"
49 #define H2FMBASE(key, radix) ((key) & ~(((hammer2_off_t)1 << (radix)) - 1))
50 #define H2FMSHIFT(radix) ((hammer2_off_t)1 << (radix))
53 * breadth-first search
55 typedef struct hammer2_chain_save {
56 TAILQ_ENTRY(hammer2_chain_save) entry;
57 hammer2_chain_t *chain;
58 int pri;
59 } hammer2_chain_save_t;
61 TAILQ_HEAD(hammer2_chain_save_list, hammer2_chain_save);
62 typedef struct hammer2_chain_save_list hammer2_chain_save_list_t;
64 typedef struct hammer2_bulkfree_info {
65 hammer2_dev_t *hmp;
66 kmem_anon_desc_t kp;
67 hammer2_off_t sbase; /* sub-loop iteration */
68 hammer2_off_t sstop;
69 hammer2_bmap_data_t *bmap;
70 int depth;
71 long count_10_00; /* staged->free */
72 long count_11_10; /* allocated->staged */
73 long count_00_11; /* (should not happen) */
74 long count_01_11; /* (should not happen) */
75 long count_10_11; /* staged->allocated */
76 long count_l0cleans;
77 long count_linadjusts;
78 long count_inodes_scanned;
79 long count_dedup_factor;
80 long bytes_scanned;
81 hammer2_off_t adj_free;
82 hammer2_tid_t mtid;
83 hammer2_tid_t saved_mirror_tid;
84 time_t save_time;
85 hammer2_chain_save_list_t list;
86 hammer2_dedup_t *dedup;
87 int pri;
88 } hammer2_bulkfree_info_t;
90 static int h2_bulkfree_test(hammer2_bulkfree_info_t *info,
91 hammer2_blockref_t *bref, int pri);
94 * General bulk scan function with callback. Called with a referenced
95 * but UNLOCKED parent. The parent is returned in the same state.
97 static
98 int
99 hammer2_bulk_scan(hammer2_chain_t *parent,
100 int (*func)(hammer2_bulkfree_info_t *info,
101 hammer2_blockref_t *bref),
102 hammer2_bulkfree_info_t *info)
104 hammer2_blockref_t bref;
105 hammer2_chain_t *chain;
106 int cache_index = -1;
107 int doabort = 0;
108 int first = 1;
110 ++info->pri;
112 hammer2_chain_lock(parent, HAMMER2_RESOLVE_ALWAYS |
113 HAMMER2_RESOLVE_SHARED);
114 chain = NULL;
117 * Generally loop on the contents if we have not been flagged
118 * for abort.
120 * Remember that these chains are completely isolated from
121 * the frontend, so we can release locks temporarily without
122 * imploding.
124 while ((doabort & HAMMER2_BULK_ABORT) == 0 &&
125 hammer2_chain_scan(parent, &chain, &bref, &first,
126 &cache_index,
127 HAMMER2_LOOKUP_NODATA |
128 HAMMER2_LOOKUP_SHARED) != NULL) {
130 * Process bref, chain is only non-NULL if the bref
131 * might be recursable (its possible that we sometimes get
132 * a non-NULL chain where the bref cannot be recursed).
134 #if 0
135 kprintf("SCAN %016jx\n", bref.data_off);
136 int xerr = tsleep(&info->pri, PCATCH, "slp", hz / 10);
137 if (xerr == EINTR || xerr == ERESTART) {
138 doabort |= HAMMER2_BULK_ABORT;
140 #endif
141 ++info->pri;
142 if (h2_bulkfree_test(info, &bref, 1))
143 continue;
145 doabort |= func(info, &bref);
147 if (doabort & HAMMER2_BULK_ABORT)
148 break;
151 * A non-null chain is always returned if it is
152 * recursive, otherwise a non-null chain might be
153 * returned but usually is not when not recursive.
155 if (chain == NULL)
156 continue;
159 * Else check type and setup depth-first scan.
161 * Account for bytes actually read.
163 info->bytes_scanned += chain->bytes;
165 switch(chain->bref.type) {
166 case HAMMER2_BREF_TYPE_INODE:
167 case HAMMER2_BREF_TYPE_FREEMAP_NODE:
168 case HAMMER2_BREF_TYPE_INDIRECT:
169 case HAMMER2_BREF_TYPE_VOLUME:
170 case HAMMER2_BREF_TYPE_FREEMAP:
171 ++info->depth;
172 if (info->depth > 16) {
173 hammer2_chain_save_t *save;
174 save = kmalloc(sizeof(*save), M_HAMMER2,
175 M_WAITOK | M_ZERO);
176 save->chain = chain;
177 hammer2_chain_ref(chain);
178 TAILQ_INSERT_TAIL(&info->list, save, entry);
180 /* guess */
181 info->pri += 10;
182 } else {
183 int savepri = info->pri;
185 hammer2_chain_unlock(chain);
186 info->pri = 0;
187 doabort |= hammer2_bulk_scan(chain, func, info);
188 info->pri += savepri;
189 hammer2_chain_lock(chain,
190 HAMMER2_RESOLVE_ALWAYS |
191 HAMMER2_RESOLVE_SHARED);
193 --info->depth;
194 break;
195 default:
196 /* does not recurse */
197 break;
200 if (chain) {
201 hammer2_chain_unlock(chain);
202 hammer2_chain_drop(chain);
206 * Save with higher pri now that we know what it is.
208 h2_bulkfree_test(info, &parent->bref, info->pri + 1);
210 hammer2_chain_unlock(parent);
212 return doabort;
216 * Bulkfree algorithm
218 * Repeat {
219 * Chain flush (partial synchronization) XXX removed
220 * Scan the whole topology - build in-memory freemap (mark 11)
221 * Reconcile the in-memory freemap against the on-disk freemap.
222 * ondisk xx -> ondisk 11 (if allocated)
223 * ondisk 11 -> ondisk 10 (if free in-memory)
224 * ondisk 10 -> ondisk 00 (if free in-memory) - on next pass
227 * The topology scan may have to be performed multiple times to window
228 * freemaps which are too large to fit in kernel memory.
230 * Races are handled using a double-transition (11->10, 10->00). The bulkfree
231 * scan snapshots the volume root's blockset and thus can run concurrent with
232 * normal operations, as long as a full flush is made between each pass to
233 * synchronize any modified chains (otherwise their blocks might be improperly
234 * freed).
236 * Temporary memory in multiples of 64KB is required to reconstruct the leaf
237 * hammer2_bmap_data blocks so they can later be compared against the live
238 * freemap. Each 64KB block represents 128 x 16KB x 1024 = ~2 GB of storage.
239 * A 32MB save area thus represents around ~1 TB. The temporary memory
240 * allocated can be specified. If it is not sufficient multiple topology
241 * passes will be made.
245 * Bulkfree callback info
247 static void cbinfo_bmap_init(hammer2_bulkfree_info_t *cbinfo, size_t size);
248 static int h2_bulkfree_callback(hammer2_bulkfree_info_t *cbinfo,
249 hammer2_blockref_t *bref);
250 static void h2_bulkfree_sync(hammer2_bulkfree_info_t *cbinfo);
251 static void h2_bulkfree_sync_adjust(hammer2_bulkfree_info_t *cbinfo,
252 hammer2_off_t data_off, hammer2_bmap_data_t *live,
253 hammer2_bmap_data_t *bmap, int nofree);
256 hammer2_bulkfree_pass(hammer2_dev_t *hmp, hammer2_ioc_bulkfree_t *bfi)
258 hammer2_bulkfree_info_t cbinfo;
259 hammer2_chain_t *vchain;
260 hammer2_chain_save_t *save;
261 hammer2_off_t incr;
262 size_t size;
263 int doabort = 0;
266 * A bulkfree operations lock is required for the duration. We
267 * must hold it across our flushes to guarantee that we never run
268 * two bulkfree passes in a row without a flush in the middle.
270 lockmgr(&hmp->bulklk, LK_EXCLUSIVE);
273 * We have to clear the live dedup cache as it might have entries
274 * that are freeable as of now. Any new entries in the dedup cache
275 * made after this point, even if they become freeable, will have
276 * previously been fully allocated and will be protected by the
277 * 2-stage bulkfree.
279 hammer2_dedup_clear(hmp);
281 #if 1
283 * XXX This has been put back in. The below check is currently
284 * disabled because it creates quite a bit of confusion, so we're
285 * going to try to fix this in a different way.
287 * XXX This has been removed. Instead of trying to flush, which
288 * appears to have a ton of races against life chains even with
289 * the two-stage scan, we simply refuse to free any blocks
290 * related to freemap chains modified after the last filesystem
291 * sync.
293 * Do a quick flush so we can snapshot vchain for any blocks that
294 * have been allocated prior to this point. We don't need to
295 * flush vnodes, logical buffers, or dirty inodes that have not
296 * allocated blocks yet. We do not want to flush the device buffers
297 * nor do we want to flush the actual volume root to disk here,
298 * that is not needed to perform the snapshot.
300 hammer2_flush_quick(hmp);
301 #endif
304 * Setup for free pass
306 bzero(&cbinfo, sizeof(cbinfo));
307 size = (bfi->size + HAMMER2_FREEMAP_LEVELN_PSIZE - 1) &
308 ~(size_t)(HAMMER2_FREEMAP_LEVELN_PSIZE - 1);
309 cbinfo.hmp = hmp;
310 cbinfo.bmap = kmem_alloc_swapbacked(&cbinfo.kp, size, VM_SUBSYS_HAMMER);
311 cbinfo.saved_mirror_tid = hmp->voldata.mirror_tid;
313 cbinfo.dedup = kmalloc(sizeof(*cbinfo.dedup) * HAMMER2_DEDUP_HEUR_SIZE,
314 M_HAMMER2, M_WAITOK | M_ZERO);
317 * Normalize start point to a 2GB boundary. We operate on a
318 * 64KB leaf bitmap boundary which represents 2GB of storage.
320 cbinfo.sbase = bfi->sbase;
321 if (cbinfo.sbase > hmp->voldata.volu_size)
322 cbinfo.sbase = hmp->voldata.volu_size;
323 cbinfo.sbase &= ~HAMMER2_FREEMAP_LEVEL1_MASK;
326 * The primary storage scan must use a snapshot of the volume
327 * root to avoid racing renames and other frontend work.
329 * Note that snapshots only snap synchronized storage, so
330 * we have to flush between each pass or we risk freeing
331 * storage allocated by the frontend.
333 vchain = hammer2_chain_bulksnap(&hmp->vchain);
334 TAILQ_INIT(&cbinfo.list);
337 * Loop on a full meta-data scan as many times as required to
338 * get through all available storage.
340 while (cbinfo.sbase < hmp->voldata.volu_size) {
342 * We have enough ram to represent (incr) bytes of storage.
343 * Each 64KB of ram represents 2GB of storage.
345 cbinfo_bmap_init(&cbinfo, size);
346 incr = size / HAMMER2_FREEMAP_LEVELN_PSIZE *
347 HAMMER2_FREEMAP_LEVEL1_SIZE;
348 if (hmp->voldata.volu_size - cbinfo.sbase < incr)
349 cbinfo.sstop = hmp->voldata.volu_size;
350 else
351 cbinfo.sstop = cbinfo.sbase + incr;
352 if (hammer2_debug & 1) {
353 kprintf("bulkfree pass %016jx/%jdGB\n",
354 (intmax_t)cbinfo.sbase,
355 (intmax_t)incr / HAMMER2_FREEMAP_LEVEL1_SIZE);
359 * Scan topology for stuff inside this range.
361 hammer2_trans_init(hmp->spmp, 0);
362 cbinfo.mtid = hammer2_trans_sub(hmp->spmp);
363 cbinfo.pri = 0;
364 doabort |= hammer2_bulk_scan(vchain, h2_bulkfree_callback,
365 &cbinfo);
367 while ((save = TAILQ_FIRST(&cbinfo.list)) != NULL &&
368 doabort == 0) {
369 TAILQ_REMOVE(&cbinfo.list, save, entry);
370 cbinfo.pri = 0;
371 doabort |= hammer2_bulk_scan(save->chain,
372 h2_bulkfree_callback,
373 &cbinfo);
374 hammer2_chain_drop(save->chain);
375 kfree(save, M_HAMMER2);
377 while (save) {
378 TAILQ_REMOVE(&cbinfo.list, save, entry);
379 hammer2_chain_drop(save->chain);
380 kfree(save, M_HAMMER2);
381 save = TAILQ_FIRST(&cbinfo.list);
384 kprintf("bulkfree lastdrop %d %d doabort=%d\n",
385 vchain->refs, vchain->core.chain_count, doabort);
388 * If complete scan succeeded we can synchronize our
389 * in-memory freemap against live storage. If an abort
390 * did occur we cannot safely synchronize our partially
391 * filled-out in-memory freemap.
393 if (doabort == 0) {
394 h2_bulkfree_sync(&cbinfo);
396 hammer2_voldata_lock(hmp);
397 hammer2_voldata_modify(hmp);
398 hmp->voldata.allocator_free += cbinfo.adj_free;
399 hammer2_voldata_unlock(hmp);
403 * Cleanup for next loop.
405 hammer2_trans_done(hmp->spmp);
406 if (doabort)
407 break;
408 cbinfo.sbase = cbinfo.sstop;
409 cbinfo.adj_free = 0;
411 hammer2_chain_bulkdrop(vchain);
412 kmem_free_swapbacked(&cbinfo.kp);
413 kfree(cbinfo.dedup, M_HAMMER2);
414 cbinfo.dedup = NULL;
416 bfi->sstop = cbinfo.sbase;
418 incr = bfi->sstop / (hmp->voldata.volu_size / 10000);
419 if (incr > 10000)
420 incr = 10000;
422 kprintf("bulkfree pass statistics (%d.%02d%% storage processed):\n",
423 (int)incr / 100,
424 (int)incr % 100);
426 kprintf(" transition->free %ld\n", cbinfo.count_10_00);
427 kprintf(" transition->staged %ld\n", cbinfo.count_11_10);
428 kprintf(" ERR(00)->allocated %ld\n", cbinfo.count_00_11);
429 kprintf(" ERR(01)->allocated %ld\n", cbinfo.count_01_11);
430 kprintf(" staged->allocated %ld\n", cbinfo.count_10_11);
431 kprintf(" ~2MB segs cleaned %ld\n", cbinfo.count_l0cleans);
432 kprintf(" linear adjusts %ld\n", cbinfo.count_linadjusts);
433 kprintf(" dedup factor %ld\n", cbinfo.count_dedup_factor);
435 lockmgr(&hmp->bulklk, LK_RELEASE);
436 /* hammer2_vfs_sync(mp, MNT_WAIT); sync needed */
438 return doabort;
441 static void
442 cbinfo_bmap_init(hammer2_bulkfree_info_t *cbinfo, size_t size)
444 hammer2_bmap_data_t *bmap = cbinfo->bmap;
445 hammer2_key_t key = cbinfo->sbase;
446 hammer2_key_t lokey;
447 hammer2_key_t hikey;
449 lokey = (cbinfo->hmp->voldata.allocator_beg + HAMMER2_SEGMASK64) &
450 ~HAMMER2_SEGMASK64;
451 hikey = cbinfo->hmp->voldata.volu_size & ~HAMMER2_SEGMASK64;
453 bzero(bmap, size);
454 while (size) {
455 if (lokey < H2FMBASE(key, HAMMER2_FREEMAP_LEVEL1_RADIX) +
456 HAMMER2_ZONE_SEG64) {
457 lokey = H2FMBASE(key, HAMMER2_FREEMAP_LEVEL1_RADIX) +
458 HAMMER2_ZONE_SEG64;
460 if (key < lokey || key >= hikey) {
461 memset(bmap->bitmapq, -1,
462 sizeof(bmap->bitmapq));
463 bmap->avail = 0;
464 bmap->linear = HAMMER2_SEGSIZE;
465 } else {
466 bmap->avail = H2FMSHIFT(HAMMER2_FREEMAP_LEVEL0_RADIX);
468 size -= sizeof(*bmap);
469 key += HAMMER2_FREEMAP_LEVEL0_SIZE;
470 ++bmap;
474 static int
475 h2_bulkfree_callback(hammer2_bulkfree_info_t *cbinfo, hammer2_blockref_t *bref)
477 hammer2_bmap_data_t *bmap;
478 hammer2_off_t data_off;
479 uint16_t class;
480 size_t bytes;
481 int radix;
482 int error;
485 * Check for signal and allow yield to userland during scan
487 if (hammer2_signal_check(&cbinfo->save_time))
488 return HAMMER2_BULK_ABORT;
489 if (bref->type == HAMMER2_BREF_TYPE_INODE) {
490 ++cbinfo->count_inodes_scanned;
491 if ((cbinfo->count_inodes_scanned & 65535) == 0)
492 kprintf(" inodes %6ld bytes %9ld\n",
493 cbinfo->count_inodes_scanned,
494 cbinfo->bytes_scanned);
498 * Calculate the data offset and determine if it is within
499 * the current freemap range being gathered.
501 error = 0;
502 data_off = bref->data_off & ~HAMMER2_OFF_MASK_RADIX;
503 if (data_off < cbinfo->sbase || data_off >= cbinfo->sstop)
504 return 0;
505 if (data_off < cbinfo->hmp->voldata.allocator_beg)
506 return 0;
507 if (data_off >= cbinfo->hmp->voldata.volu_size)
508 return 0;
511 * Calculate the information needed to generate the in-memory
512 * freemap record.
514 * Hammer2 does not allow allocations to cross the L1 (2GB) boundary,
515 * it's a problem if it does. (Or L0 (2MB) for that matter).
517 radix = (int)(bref->data_off & HAMMER2_OFF_MASK_RADIX);
518 bytes = (size_t)1 << radix;
519 class = (bref->type << 8) | hammer2_devblkradix(radix);
521 if (data_off + bytes >= cbinfo->sstop) {
522 kprintf("hammer2_bulkfree_scan: illegal 2GB boundary "
523 "%016jx %016jx/%d\n",
524 (intmax_t)bref->data_off,
525 (intmax_t)bref->key,
526 bref->keybits);
527 bytes = cbinfo->sstop - data_off; /* XXX */
531 * Convert to a storage offset relative to the beginning of the
532 * storage range we are collecting. Then lookup the level0 bmap entry.
534 data_off -= cbinfo->sbase;
535 bmap = cbinfo->bmap + (data_off >> HAMMER2_FREEMAP_LEVEL0_RADIX);
538 * Convert data_off to a bmap-relative value (~2MB storage range).
539 * Adjust linear, class, and avail.
541 * Hammer2 does not allow allocations to cross the L0 (2MB) boundary,
543 data_off &= HAMMER2_FREEMAP_LEVEL0_MASK;
544 if (data_off + bytes > HAMMER2_FREEMAP_LEVEL0_SIZE) {
545 kprintf("hammer2_bulkfree_scan: illegal 2MB boundary "
546 "%016jx %016jx/%d\n",
547 (intmax_t)bref->data_off,
548 (intmax_t)bref->key,
549 bref->keybits);
550 bytes = HAMMER2_FREEMAP_LEVEL0_SIZE - data_off;
553 if (bmap->class == 0) {
554 bmap->class = class;
555 bmap->avail = HAMMER2_FREEMAP_LEVEL0_SIZE;
557 if (bmap->class != class) {
558 kprintf("hammer2_bulkfree_scan: illegal mixed class "
559 "%016jx %016jx/%d (%04x vs %04x)\n",
560 (intmax_t)bref->data_off,
561 (intmax_t)bref->key,
562 bref->keybits,
563 class, bmap->class);
565 if (bmap->linear < (int32_t)data_off + (int32_t)bytes)
566 bmap->linear = (int32_t)data_off + (int32_t)bytes;
569 * Adjust the hammer2_bitmap_t bitmap[HAMMER2_BMAP_ELEMENTS].
570 * 64-bit entries, 2 bits per entry, to code 11.
572 * NOTE: The allocation can be smaller than HAMMER2_FREEMAP_BLOCK_SIZE.
574 while (bytes > 0) {
575 int bindex;
576 hammer2_bitmap_t bmask;
578 bindex = (int)data_off >> (HAMMER2_FREEMAP_BLOCK_RADIX +
579 HAMMER2_BMAP_INDEX_RADIX);
580 bmask = (hammer2_bitmap_t)3 <<
581 ((((int)data_off & HAMMER2_BMAP_INDEX_MASK) >>
582 HAMMER2_FREEMAP_BLOCK_RADIX) << 1);
585 * NOTE! The (avail) calculation is bitmap-granular. Multiple
586 * sub-granular records can wind up at the same bitmap
587 * position.
589 if ((bmap->bitmapq[bindex] & bmask) == 0) {
590 if (bytes < HAMMER2_FREEMAP_BLOCK_SIZE) {
591 bmap->avail -= HAMMER2_FREEMAP_BLOCK_SIZE;
592 } else {
593 bmap->avail -= bytes;
595 bmap->bitmapq[bindex] |= bmask;
597 data_off += HAMMER2_FREEMAP_BLOCK_SIZE;
598 if (bytes < HAMMER2_FREEMAP_BLOCK_SIZE)
599 bytes = 0;
600 else
601 bytes -= HAMMER2_FREEMAP_BLOCK_SIZE;
603 return error;
607 * Synchronize the in-memory bitmap with the live freemap. This is not a
608 * direct copy. Instead the bitmaps must be compared:
610 * In-memory Live-freemap
611 * 00 11 -> 10 (do nothing if live modified)
612 * 10 -> 00 (do nothing if live modified)
613 * 11 10 -> 11 handles race against live
614 * ** -> 11 nominally warn of corruption
617 static void
618 h2_bulkfree_sync(hammer2_bulkfree_info_t *cbinfo)
620 hammer2_off_t data_off;
621 hammer2_key_t key;
622 hammer2_key_t key_dummy;
623 hammer2_bmap_data_t *bmap;
624 hammer2_bmap_data_t *live;
625 hammer2_chain_t *live_parent;
626 hammer2_chain_t *live_chain;
627 int cache_index = -1;
628 int bmapindex;
630 kprintf("hammer2_bulkfree - range ");
632 if (cbinfo->sbase < cbinfo->hmp->voldata.allocator_beg)
633 kprintf("%016jx-",
634 (intmax_t)cbinfo->hmp->voldata.allocator_beg);
635 else
636 kprintf("%016jx-",
637 (intmax_t)cbinfo->sbase);
639 if (cbinfo->sstop > cbinfo->hmp->voldata.volu_size)
640 kprintf("%016jx\n",
641 (intmax_t)cbinfo->hmp->voldata.volu_size);
642 else
643 kprintf("%016jx\n",
644 (intmax_t)cbinfo->sstop);
646 data_off = cbinfo->sbase;
647 bmap = cbinfo->bmap;
649 live_parent = &cbinfo->hmp->fchain;
650 hammer2_chain_ref(live_parent);
651 hammer2_chain_lock(live_parent, HAMMER2_RESOLVE_ALWAYS);
652 live_chain = NULL;
655 * Iterate each hammer2_bmap_data_t line (128 bytes) managing
656 * 4MB of storage.
658 while (data_off < cbinfo->sstop) {
660 * The freemap is not used below allocator_beg or beyond
661 * volu_size.
663 int nofree;
665 if (data_off < cbinfo->hmp->voldata.allocator_beg)
666 goto next;
667 if (data_off >= cbinfo->hmp->voldata.volu_size)
668 goto next;
671 * Locate the freemap leaf on the live filesystem
673 key = (data_off & ~HAMMER2_FREEMAP_LEVEL1_MASK);
674 nofree = 0;
676 if (live_chain == NULL || live_chain->bref.key != key) {
677 if (live_chain) {
678 hammer2_chain_unlock(live_chain);
679 hammer2_chain_drop(live_chain);
681 live_chain = hammer2_chain_lookup(
682 &live_parent,
683 &key_dummy,
684 key,
685 key + HAMMER2_FREEMAP_LEVEL1_MASK,
686 &cache_index,
687 HAMMER2_LOOKUP_ALWAYS);
689 #if 0
691 * If recent allocations were made we avoid races by
692 * not staging or freeing any blocks. We can still
693 * remark blocks as fully allocated.
695 if (live_chain) {
696 if (hammer2_debug & 1) {
697 kprintf("live_chain %016jx\n",
698 (intmax_t)key);
700 if (live_chain->bref.mirror_tid >
701 cbinfo->saved_mirror_tid) {
702 kprintf("hammer2_bulkfree: "
703 "avoid %016jx\n",
704 data_off);
705 nofree = 1;
706 } else {
707 nofree = 0;
710 #endif
712 if (live_chain == NULL) {
714 * XXX if we implement a full recovery mode we need
715 * to create/recreate missing freemap chains if our
716 * bmap has any allocated blocks.
718 if (bmap->class &&
719 bmap->avail != HAMMER2_FREEMAP_LEVEL0_SIZE) {
720 kprintf("hammer2_bulkfree: cannot locate "
721 "live leaf for allocated data "
722 "near %016jx\n",
723 (intmax_t)data_off);
725 goto next;
727 if (live_chain->error) {
728 kprintf("hammer2_bulkfree: error %s looking up "
729 "live leaf for allocated data near %016jx\n",
730 hammer2_error_str(live_chain->error),
731 (intmax_t)data_off);
732 hammer2_chain_unlock(live_chain);
733 hammer2_chain_drop(live_chain);
734 live_chain = NULL;
735 goto next;
738 bmapindex = (data_off & HAMMER2_FREEMAP_LEVEL1_MASK) >>
739 HAMMER2_FREEMAP_LEVEL0_RADIX;
740 live = &live_chain->data->bmdata[bmapindex];
743 * TODO - we could shortcut this by testing that both
744 * live->class and bmap->class are 0, and both avails are
745 * set to HAMMER2_FREEMAP_LEVEL0_SIZE (4MB).
747 if (bcmp(live->bitmapq, bmap->bitmapq,
748 sizeof(bmap->bitmapq)) == 0) {
749 goto next;
751 if (hammer2_debug & 1) {
752 kprintf("live %016jx %04d.%04x (avail=%d)\n",
753 data_off, bmapindex, live->class, live->avail);
756 hammer2_chain_modify(live_chain, cbinfo->mtid, 0, 0);
757 live = &live_chain->data->bmdata[bmapindex];
759 h2_bulkfree_sync_adjust(cbinfo, data_off, live, bmap, nofree);
760 next:
761 data_off += HAMMER2_FREEMAP_LEVEL0_SIZE;
762 ++bmap;
764 if (live_chain) {
765 hammer2_chain_unlock(live_chain);
766 hammer2_chain_drop(live_chain);
768 if (live_parent) {
769 hammer2_chain_unlock(live_parent);
770 hammer2_chain_drop(live_parent);
775 * When bulkfree is finally able to free a block it must make sure that
776 * the INVALOK bit in any cached DIO is cleared prior to the block being
777 * reused.
779 static
780 void
781 fixup_dio(hammer2_dev_t *hmp, hammer2_off_t data_off, int bindex, int scount)
783 data_off += (scount >> 1) * HAMMER2_FREEMAP_BLOCK_SIZE;
784 data_off += bindex *
785 (HAMMER2_FREEMAP_BLOCK_SIZE * HAMMER2_BMAP_BLOCKS_PER_ELEMENT);
786 hammer2_io_resetinval(hmp, data_off);
790 * Merge the bulkfree bitmap against the existing bitmap.
792 * If nofree is non-zero the merge will only mark free blocks as allocated
793 * and will refuse to free any blocks.
795 static
796 void
797 h2_bulkfree_sync_adjust(hammer2_bulkfree_info_t *cbinfo,
798 hammer2_off_t data_off, hammer2_bmap_data_t *live,
799 hammer2_bmap_data_t *bmap, int nofree)
801 int bindex;
802 int scount;
803 hammer2_bitmap_t lmask;
804 hammer2_bitmap_t mmask;
806 for (bindex = 0; bindex < HAMMER2_BMAP_ELEMENTS; ++bindex) {
807 lmask = live->bitmapq[bindex]; /* live */
808 mmask = bmap->bitmapq[bindex]; /* snapshotted bulkfree */
809 if (lmask == mmask)
810 continue;
812 for (scount = 0;
813 scount < HAMMER2_BMAP_BITS_PER_ELEMENT;
814 scount += 2) {
815 if ((mmask & 3) == 0) {
817 * in-memory 00 live 11 -> 10
818 * live 10 -> 00
820 * Storage might be marked allocated or
821 * staged and must be remarked staged or
822 * free.
824 switch (lmask & 3) {
825 case 0: /* 00 */
826 break;
827 case 1: /* 01 */
828 kprintf("hammer2_bulkfree: cannot "
829 "transition m=00/l=01\n");
830 break;
831 case 2: /* 10 -> 00 */
832 if (nofree)
833 break;
834 live->bitmapq[bindex] &=
835 ~((hammer2_bitmap_t)2 << scount);
836 live->avail +=
837 HAMMER2_FREEMAP_BLOCK_SIZE;
838 if (live->avail >
839 HAMMER2_FREEMAP_LEVEL0_SIZE) {
840 live->avail =
841 HAMMER2_FREEMAP_LEVEL0_SIZE;
843 cbinfo->adj_free +=
844 HAMMER2_FREEMAP_BLOCK_SIZE;
845 ++cbinfo->count_10_00;
846 break;
847 case 3: /* 11 -> 10 */
848 if (nofree)
849 break;
850 live->bitmapq[bindex] &=
851 ~((hammer2_bitmap_t)1 << scount);
852 ++cbinfo->count_11_10;
853 fixup_dio(cbinfo->hmp, data_off,
854 bindex, scount);
855 break;
857 } else if ((mmask & 3) == 3) {
859 * in-memory 11 live 10 -> 11
860 * live ** -> 11
862 * Storage might be incorrectly marked free
863 * or staged and must be remarked fully
864 * allocated.
866 switch (lmask & 3) {
867 case 0: /* 00 */
868 ++cbinfo->count_00_11;
869 cbinfo->adj_free -=
870 HAMMER2_FREEMAP_BLOCK_SIZE;
871 live->avail -=
872 HAMMER2_FREEMAP_BLOCK_SIZE;
873 if ((int32_t)live->avail < 0)
874 live->avail = 0;
875 break;
876 case 1: /* 01 */
877 ++cbinfo->count_01_11;
878 break;
879 case 2: /* 10 -> 11 */
880 ++cbinfo->count_10_11;
881 break;
882 case 3: /* 11 */
883 break;
885 live->bitmapq[bindex] |=
886 ((hammer2_bitmap_t)3 << scount);
888 mmask >>= 2;
889 lmask >>= 2;
894 * Determine if the live bitmap is completely free and reset its
895 * fields if so. Otherwise check to see if we can reduce the linear
896 * offset.
898 for (bindex = HAMMER2_BMAP_ELEMENTS - 1; bindex >= 0; --bindex) {
899 if (live->bitmapq[bindex] != 0)
900 break;
902 if (nofree) {
903 /* do nothing */
904 } else if (bindex < 0) {
905 live->avail = HAMMER2_FREEMAP_LEVEL0_SIZE;
906 live->class = 0;
907 live->linear = 0;
908 ++cbinfo->count_l0cleans;
909 } else if (bindex < 7) {
910 ++bindex;
911 if (live->linear > bindex * HAMMER2_FREEMAP_BLOCK_SIZE) {
912 live->linear = bindex * HAMMER2_FREEMAP_BLOCK_SIZE;
913 ++cbinfo->count_linadjusts;
917 * XXX this fine-grained measure still has some issues.
919 if (live->linear < bindex * HAMMER2_FREEMAP_BLOCK_SIZE) {
920 live->linear = bindex * HAMMER2_FREEMAP_BLOCK_SIZE;
921 ++cbinfo->count_linadjusts;
923 } else {
924 live->linear = HAMMER2_SEGSIZE;
927 #if 0
928 if (bmap->class) {
929 kprintf("%016jx %04d.%04x (avail=%7d) "
930 "%08x %08x %08x %08x %08x %08x %08x %08x\n",
931 (intmax_t)data_off,
932 (int)((data_off &
933 HAMMER2_FREEMAP_LEVEL1_MASK) >>
934 HAMMER2_FREEMAP_LEVEL0_RADIX),
935 bmap->class,
936 bmap->avail,
937 bmap->bitmap[0], bmap->bitmap[1],
938 bmap->bitmap[2], bmap->bitmap[3],
939 bmap->bitmap[4], bmap->bitmap[5],
940 bmap->bitmap[6], bmap->bitmap[7]);
942 #endif
946 * BULKFREE DEDUP HEURISTIC
948 * WARNING! This code is SMP safe but the heuristic allows SMP collisions.
949 * All fields must be loaded into locals and validated.
951 static
953 h2_bulkfree_test(hammer2_bulkfree_info_t *cbinfo, hammer2_blockref_t *bref,
954 int pri)
956 hammer2_dedup_t *dedup;
957 int best;
958 int n;
959 int i;
961 n = hammer2_icrc32(&bref->data_off, sizeof(bref->data_off));
962 dedup = cbinfo->dedup + (n & (HAMMER2_DEDUP_HEUR_MASK & ~7));
964 for (i = best = 0; i < 8; ++i) {
965 if (dedup[i].data_off == bref->data_off) {
966 if (dedup[i].ticks < pri)
967 dedup[i].ticks = pri;
968 if (pri == 1)
969 cbinfo->count_dedup_factor += dedup[i].ticks;
970 return 1;
972 if (dedup[i].ticks < dedup[best].ticks)
973 best = i;
975 dedup[best].data_off = bref->data_off;
976 dedup[best].ticks = pri;
978 return 0;