1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
|
/*
* preallocate filter driver
*
* The driver performs preallocate operation: it is injected above
* some node, and before each write over EOF it does additional preallocating
* write-zeroes request.
*
* Copyright (c) 2020 Virtuozzo International GmbH.
*
* Author:
* Sementsov-Ogievskiy Vladimir <vsementsov@virtuozzo.com>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "qemu/osdep.h"
#include "qapi/error.h"
#include "qemu/module.h"
#include "qemu/option.h"
#include "qemu/units.h"
#include "block/block-io.h"
#include "block/block_int.h"
typedef struct PreallocateOpts {
int64_t prealloc_size;
int64_t prealloc_align;
} PreallocateOpts;
typedef struct BDRVPreallocateState {
PreallocateOpts opts;
/*
* Track real data end, to crop preallocation on close. If < 0 the status is
* unknown.
*
* @data_end is a maximum of file size on open (or when we get write/resize
* permissions) and all write request ends after it. So it's safe to
* truncate to data_end if it is valid.
*/
int64_t data_end;
/*
* Start of trailing preallocated area which reads as zero. May be smaller
* than data_end, if user does over-EOF write zero operation. If < 0 the
* status is unknown.
*
* If both @zero_start and @file_end are valid, the region
* [@zero_start, @file_end) is known to be preallocated zeroes. If @file_end
* is not valid, @zero_start doesn't make much sense.
*/
int64_t zero_start;
/*
* Real end of file. Actually the cache for bdrv_getlength(bs->file->bs),
* to avoid extra lseek() calls on each write operation. If < 0 the status
* is unknown.
*/
int64_t file_end;
/*
* All three states @data_end, @zero_start and @file_end are guaranteed to
* be invalid (< 0) when we don't have both exclusive BLK_PERM_RESIZE and
* BLK_PERM_WRITE permissions on file child.
*/
/* Gives up the resize permission on children when parents don't need it */
QEMUBH *drop_resize_bh;
} BDRVPreallocateState;
static int preallocate_drop_resize(BlockDriverState *bs, Error **errp);
static void preallocate_drop_resize_bh(void *opaque);
#define PREALLOCATE_OPT_PREALLOC_ALIGN "prealloc-align"
#define PREALLOCATE_OPT_PREALLOC_SIZE "prealloc-size"
static QemuOptsList runtime_opts = {
.name = "preallocate",
.head = QTAILQ_HEAD_INITIALIZER(runtime_opts.head),
.desc = {
{
.name = PREALLOCATE_OPT_PREALLOC_ALIGN,
.type = QEMU_OPT_SIZE,
.help = "on preallocation, align file length to this number, "
"default 1M",
},
{
.name = PREALLOCATE_OPT_PREALLOC_SIZE,
.type = QEMU_OPT_SIZE,
.help = "how much to preallocate, default 128M",
},
{ /* end of list */ }
},
};
static bool preallocate_absorb_opts(PreallocateOpts *dest, QDict *options,
BlockDriverState *child_bs, Error **errp)
{
QemuOpts *opts = qemu_opts_create(&runtime_opts, NULL, 0, &error_abort);
if (!qemu_opts_absorb_qdict(opts, options, errp)) {
return false;
}
dest->prealloc_align =
qemu_opt_get_size(opts, PREALLOCATE_OPT_PREALLOC_ALIGN, 1 * MiB);
dest->prealloc_size =
qemu_opt_get_size(opts, PREALLOCATE_OPT_PREALLOC_SIZE, 128 * MiB);
qemu_opts_del(opts);
if (!QEMU_IS_ALIGNED(dest->prealloc_align, BDRV_SECTOR_SIZE)) {
error_setg(errp, "prealloc-align parameter of preallocate filter "
"is not aligned to %llu", BDRV_SECTOR_SIZE);
return false;
}
if (!QEMU_IS_ALIGNED(dest->prealloc_align,
child_bs->bl.request_alignment)) {
error_setg(errp, "prealloc-align parameter of preallocate filter "
"is not aligned to underlying node request alignment "
"(%" PRIi32 ")", child_bs->bl.request_alignment);
return false;
}
return true;
}
static int preallocate_open(BlockDriverState *bs, QDict *options, int flags,
Error **errp)
{
BDRVPreallocateState *s = bs->opaque;
int ret;
GLOBAL_STATE_CODE();
/*
* s->data_end and friends should be initialized on permission update.
* For this to work, mark them invalid.
*/
s->file_end = s->zero_start = s->data_end = -EINVAL;
s->drop_resize_bh = qemu_bh_new(preallocate_drop_resize_bh, bs);
ret = bdrv_open_file_child(NULL, options, "file", bs, errp);
if (ret < 0) {
return ret;
}
GRAPH_RDLOCK_GUARD_MAINLOOP();
if (!preallocate_absorb_opts(&s->opts, options, bs->file->bs, errp)) {
return -EINVAL;
}
bs->supported_write_flags = BDRV_REQ_WRITE_UNCHANGED |
(BDRV_REQ_FUA & bs->file->bs->supported_write_flags);
bs->supported_zero_flags = BDRV_REQ_WRITE_UNCHANGED |
((BDRV_REQ_FUA | BDRV_REQ_MAY_UNMAP | BDRV_REQ_NO_FALLBACK) &
bs->file->bs->supported_zero_flags);
return 0;
}
static int GRAPH_RDLOCK
preallocate_truncate_to_real_size(BlockDriverState *bs, Error **errp)
{
BDRVPreallocateState *s = bs->opaque;
int ret;
if (s->file_end < 0) {
s->file_end = bdrv_getlength(bs->file->bs);
if (s->file_end < 0) {
error_setg_errno(errp, -s->file_end, "Failed to get file length");
return s->file_end;
}
}
if (s->data_end < s->file_end) {
ret = bdrv_truncate(bs->file, s->data_end, true, PREALLOC_MODE_OFF, 0,
NULL);
if (ret < 0) {
error_setg_errno(errp, -ret, "Failed to drop preallocation");
s->file_end = ret;
return ret;
}
s->file_end = s->data_end;
}
return 0;
}
static void preallocate_close(BlockDriverState *bs)
{
BDRVPreallocateState *s = bs->opaque;
GLOBAL_STATE_CODE();
GRAPH_RDLOCK_GUARD_MAINLOOP();
qemu_bh_cancel(s->drop_resize_bh);
qemu_bh_delete(s->drop_resize_bh);
if (s->data_end >= 0) {
preallocate_truncate_to_real_size(bs, NULL);
}
}
/*
* Handle reopen.
*
* We must implement reopen handlers, otherwise reopen just don't work. Handle
* new options and don't care about preallocation state, as it is handled in
* set/check permission handlers.
*/
static int preallocate_reopen_prepare(BDRVReopenState *reopen_state,
BlockReopenQueue *queue, Error **errp)
{
PreallocateOpts *opts = g_new0(PreallocateOpts, 1);
int ret;
GLOBAL_STATE_CODE();
GRAPH_RDLOCK_GUARD_MAINLOOP();
if (!preallocate_absorb_opts(opts, reopen_state->options,
reopen_state->bs->file->bs, errp)) {
g_free(opts);
return -EINVAL;
}
/*
* Drop the preallocation already here if reopening read-only. The child
* might also be reopened read-only and then scheduling a BH during the
* permission update is too late.
*/
if ((reopen_state->flags & BDRV_O_RDWR) == 0) {
ret = preallocate_drop_resize(reopen_state->bs, errp);
if (ret < 0) {
g_free(opts);
return ret;
}
}
reopen_state->opaque = opts;
return 0;
}
static void preallocate_reopen_commit(BDRVReopenState *state)
{
BDRVPreallocateState *s = state->bs->opaque;
s->opts = *(PreallocateOpts *)state->opaque;
g_free(state->opaque);
state->opaque = NULL;
}
static void preallocate_reopen_abort(BDRVReopenState *state)
{
g_free(state->opaque);
state->opaque = NULL;
}
static int coroutine_fn GRAPH_RDLOCK
preallocate_co_preadv_part(BlockDriverState *bs, int64_t offset, int64_t bytes,
QEMUIOVector *qiov, size_t qiov_offset,
BdrvRequestFlags flags)
{
return bdrv_co_preadv_part(bs->file, offset, bytes, qiov, qiov_offset,
flags);
}
static int coroutine_fn GRAPH_RDLOCK
preallocate_co_pdiscard(BlockDriverState *bs, int64_t offset, int64_t bytes)
{
return bdrv_co_pdiscard(bs->file, offset, bytes);
}
static bool can_write_resize(uint64_t perm)
{
return (perm & BLK_PERM_WRITE) && (perm & BLK_PERM_RESIZE);
}
static bool GRAPH_RDLOCK has_prealloc_perms(BlockDriverState *bs)
{
BDRVPreallocateState *s = bs->opaque;
if (can_write_resize(bs->file->perm)) {
assert(!(bs->file->shared_perm & BLK_PERM_WRITE));
assert(!(bs->file->shared_perm & BLK_PERM_RESIZE));
return true;
}
assert(s->data_end < 0);
assert(s->zero_start < 0);
assert(s->file_end < 0);
return false;
}
/*
* Call on each write. Returns true if @want_merge_zero is true and the region
* [offset, offset + bytes) is zeroed (as a result of this call or earlier
* preallocation).
*
* want_merge_zero is used to merge write-zero request with preallocation in
* one bdrv_co_pwrite_zeroes() call.
*/
static bool coroutine_fn GRAPH_RDLOCK
handle_write(BlockDriverState *bs, int64_t offset, int64_t bytes,
bool want_merge_zero)
{
BDRVPreallocateState *s = bs->opaque;
int64_t end = offset + bytes;
int64_t prealloc_start, prealloc_end;
int ret;
uint32_t file_align = bs->file->bs->bl.request_alignment;
uint32_t prealloc_align = MAX(s->opts.prealloc_align, file_align);
assert(QEMU_IS_ALIGNED(prealloc_align, file_align));
if (!has_prealloc_perms(bs)) {
/* We don't have state neither should try to recover it */
return false;
}
if (s->data_end < 0) {
s->data_end = bdrv_co_getlength(bs->file->bs);
if (s->data_end < 0) {
return false;
}
if (s->file_end < 0) {
s->file_end = s->data_end;
}
}
if (end <= s->data_end) {
return false;
}
/* We have valid s->data_end, and request writes beyond it. */
s->data_end = end;
if (s->zero_start < 0 || !want_merge_zero) {
s->zero_start = end;
}
if (s->file_end < 0) {
s->file_end = bdrv_co_getlength(bs->file->bs);
if (s->file_end < 0) {
return false;
}
}
/* Now s->data_end, s->zero_start and s->file_end are valid. */
if (end <= s->file_end) {
/* No preallocation needed. */
return want_merge_zero && offset >= s->zero_start;
}
/* Now we want new preallocation, as request writes beyond s->file_end. */
prealloc_start = QEMU_ALIGN_UP(
want_merge_zero ? MIN(offset, s->file_end) : s->file_end,
file_align);
prealloc_end = QEMU_ALIGN_UP(
MAX(prealloc_start, end) + s->opts.prealloc_size,
prealloc_align);
want_merge_zero = want_merge_zero && (prealloc_start <= offset);
ret = bdrv_co_pwrite_zeroes(
bs->file, prealloc_start, prealloc_end - prealloc_start,
BDRV_REQ_NO_FALLBACK | BDRV_REQ_SERIALISING | BDRV_REQ_NO_WAIT);
if (ret < 0) {
s->file_end = ret;
return false;
}
s->file_end = prealloc_end;
return want_merge_zero;
}
static int coroutine_fn GRAPH_RDLOCK
preallocate_co_pwrite_zeroes(BlockDriverState *bs, int64_t offset,
int64_t bytes, BdrvRequestFlags flags)
{
bool want_merge_zero =
!(flags & ~(BDRV_REQ_ZERO_WRITE | BDRV_REQ_NO_FALLBACK));
if (handle_write(bs, offset, bytes, want_merge_zero)) {
return 0;
}
return bdrv_co_pwrite_zeroes(bs->file, offset, bytes, flags);
}
static int coroutine_fn GRAPH_RDLOCK
preallocate_co_pwritev_part(BlockDriverState *bs, int64_t offset, int64_t bytes,
QEMUIOVector *qiov, size_t qiov_offset,
BdrvRequestFlags flags)
{
handle_write(bs, offset, bytes, false);
return bdrv_co_pwritev_part(bs->file, offset, bytes, qiov, qiov_offset,
flags);
}
static int coroutine_fn GRAPH_RDLOCK
preallocate_co_truncate(BlockDriverState *bs, int64_t offset,
bool exact, PreallocMode prealloc,
BdrvRequestFlags flags, Error **errp)
{
ERRP_GUARD();
BDRVPreallocateState *s = bs->opaque;
int ret;
if (s->data_end >= 0 && offset > s->data_end) {
if (s->file_end < 0) {
s->file_end = bdrv_co_getlength(bs->file->bs);
if (s->file_end < 0) {
error_setg(errp, "failed to get file length");
return s->file_end;
}
}
if (prealloc == PREALLOC_MODE_FALLOC) {
/*
* If offset <= s->file_end, the task is already done, just
* update s->data_end, to move part of "filter preallocation"
* to "preallocation requested by user".
* Otherwise just proceed to preallocate missing part.
*/
if (offset <= s->file_end) {
s->data_end = offset;
return 0;
}
} else {
/*
* We have to drop our preallocation, to
* - avoid "Cannot use preallocation for shrinking files" in
* case of offset < file_end
* - give PREALLOC_MODE_OFF a chance to keep small disk
* usage
* - give PREALLOC_MODE_FULL a chance to actually write the
* whole region as user expects
*/
if (s->file_end > s->data_end) {
ret = bdrv_co_truncate(bs->file, s->data_end, true,
PREALLOC_MODE_OFF, 0, errp);
if (ret < 0) {
s->file_end = ret;
error_prepend(errp, "preallocate-filter: failed to drop "
"write-zero preallocation: ");
return ret;
}
s->file_end = s->data_end;
}
}
s->data_end = offset;
}
ret = bdrv_co_truncate(bs->file, offset, exact, prealloc, flags, errp);
if (ret < 0) {
s->file_end = s->zero_start = s->data_end = ret;
return ret;
}
if (has_prealloc_perms(bs)) {
s->file_end = s->zero_start = s->data_end = offset;
}
return 0;
}
static int coroutine_fn GRAPH_RDLOCK preallocate_co_flush(BlockDriverState *bs)
{
return bdrv_co_flush(bs->file->bs);
}
static int64_t coroutine_fn GRAPH_RDLOCK
preallocate_co_getlength(BlockDriverState *bs)
{
int64_t ret;
BDRVPreallocateState *s = bs->opaque;
if (s->data_end >= 0) {
return s->data_end;
}
ret = bdrv_co_getlength(bs->file->bs);
if (has_prealloc_perms(bs)) {
s->file_end = s->zero_start = s->data_end = ret;
}
return ret;
}
static int GRAPH_RDLOCK
preallocate_drop_resize(BlockDriverState *bs, Error **errp)
{
BDRVPreallocateState *s = bs->opaque;
int ret;
if (s->data_end < 0) {
return 0;
}
/*
* Before switching children to be read-only, truncate them to remove
* the preallocation and let them have the real size.
*/
ret = preallocate_truncate_to_real_size(bs, errp);
if (ret < 0) {
return ret;
}
/*
* We'll drop our permissions and will allow other users to take write and
* resize permissions (see preallocate_child_perm). Anyone will be able to
* change the child, so mark all states invalid. We'll regain control if a
* parent requests write access again.
*/
s->data_end = s->file_end = s->zero_start = -EINVAL;
bdrv_child_refresh_perms(bs, bs->file, NULL);
return 0;
}
static void preallocate_drop_resize_bh(void *opaque)
{
GLOBAL_STATE_CODE();
GRAPH_RDLOCK_GUARD_MAINLOOP();
/*
* In case of errors, we'll simply keep the exclusive lock on the image
* indefinitely.
*/
preallocate_drop_resize(opaque, NULL);
}
static void GRAPH_RDLOCK
preallocate_set_perm(BlockDriverState *bs, uint64_t perm, uint64_t shared)
{
BDRVPreallocateState *s = bs->opaque;
if (can_write_resize(perm)) {
qemu_bh_cancel(s->drop_resize_bh);
if (s->data_end < 0) {
s->data_end = s->file_end = s->zero_start =
bs->file->bs->total_sectors * BDRV_SECTOR_SIZE;
}
} else {
qemu_bh_schedule(s->drop_resize_bh);
}
}
static void preallocate_child_perm(BlockDriverState *bs, BdrvChild *c,
BdrvChildRole role, BlockReopenQueue *reopen_queue,
uint64_t perm, uint64_t shared, uint64_t *nperm, uint64_t *nshared)
{
BDRVPreallocateState *s = bs->opaque;
bdrv_default_perms(bs, c, role, reopen_queue, perm, shared, nperm, nshared);
/*
* We need exclusive write and resize permissions on the child not only when
* the parent can write to it, but also after the parent gave up write
* permissions until preallocate_drop_resize() has completed.
*/
if (can_write_resize(perm) || s->data_end != -EINVAL) {
*nperm |= BLK_PERM_WRITE | BLK_PERM_RESIZE;
/*
* Don't share, to keep our states s->file_end, s->data_end and
* s->zero_start valid.
*/
*nshared &= ~(BLK_PERM_WRITE | BLK_PERM_RESIZE);
}
}
static BlockDriver bdrv_preallocate_filter = {
.format_name = "preallocate",
.instance_size = sizeof(BDRVPreallocateState),
.bdrv_co_getlength = preallocate_co_getlength,
.bdrv_open = preallocate_open,
.bdrv_close = preallocate_close,
.bdrv_reopen_prepare = preallocate_reopen_prepare,
.bdrv_reopen_commit = preallocate_reopen_commit,
.bdrv_reopen_abort = preallocate_reopen_abort,
.bdrv_co_preadv_part = preallocate_co_preadv_part,
.bdrv_co_pwritev_part = preallocate_co_pwritev_part,
.bdrv_co_pwrite_zeroes = preallocate_co_pwrite_zeroes,
.bdrv_co_pdiscard = preallocate_co_pdiscard,
.bdrv_co_flush = preallocate_co_flush,
.bdrv_co_truncate = preallocate_co_truncate,
.bdrv_set_perm = preallocate_set_perm,
.bdrv_child_perm = preallocate_child_perm,
.is_filter = true,
};
static void bdrv_preallocate_init(void)
{
bdrv_register(&bdrv_preallocate_filter);
}
block_init(bdrv_preallocate_init);
|