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
|
// Copyright (C) 2022-2023 Luke Shumaker <lukeshu@lukeshu.com>
//
// SPDX-License-Identifier: GPL-2.0-or-later
package btrfsvol
import (
"bytes"
"errors"
"fmt"
"os"
"reflect"
"github.com/datawire/dlib/derror"
"git.lukeshu.com/btrfs-progs-ng/lib/containers"
"git.lukeshu.com/btrfs-progs-ng/lib/diskio"
)
type LogicalVolume[PhysicalVolume diskio.File[PhysicalAddr]] struct {
name string
id2pv map[DeviceID]PhysicalVolume
logical2physical *containers.RBTree[chunkMapping]
physical2logical map[DeviceID]*containers.RBTree[devextMapping]
}
var _ diskio.File[LogicalAddr] = (*LogicalVolume[diskio.File[PhysicalAddr]])(nil)
func (lv *LogicalVolume[PhysicalVolume]) init() {
if lv.id2pv == nil {
lv.id2pv = make(map[DeviceID]PhysicalVolume)
}
if lv.logical2physical == nil {
lv.logical2physical = new(containers.RBTree[chunkMapping])
}
if lv.physical2logical == nil {
lv.physical2logical = make(map[DeviceID]*containers.RBTree[devextMapping], len(lv.id2pv))
}
for devid := range lv.id2pv {
if _, ok := lv.physical2logical[devid]; !ok {
lv.physical2logical[devid] = new(containers.RBTree[devextMapping])
}
}
}
func (lv *LogicalVolume[PhysicalVolume]) SetName(name string) {
lv.name = name
}
func (lv *LogicalVolume[PhysicalVolume]) Name() string {
return lv.name
}
func (lv *LogicalVolume[PhysicalVolume]) Size() LogicalAddr {
lv.init()
lastChunk := lv.logical2physical.Max()
if lastChunk == nil {
return 0
}
return lastChunk.Value.LAddr.Add(lastChunk.Value.Size)
}
func (lv *LogicalVolume[PhysicalVolume]) Close() error {
var errs derror.MultiError
for _, dev := range lv.id2pv {
if err := dev.Close(); err != nil && err == nil {
errs = append(errs, err)
}
}
if errs != nil {
return errs
}
return nil
}
func (lv *LogicalVolume[PhysicalVolume]) AddPhysicalVolume(id DeviceID, dev PhysicalVolume) error {
lv.init()
if other, exists := lv.id2pv[id]; exists {
return fmt.Errorf("(%p).AddPhysicalVolume: cannot add physical volume %q: already have physical volume %q with id=%v",
lv, dev.Name(), other.Name(), id)
}
lv.id2pv[id] = dev
lv.physical2logical[id] = new(containers.RBTree[devextMapping])
return nil
}
func (lv *LogicalVolume[PhysicalVolume]) PhysicalVolumes() map[DeviceID]PhysicalVolume {
dup := make(map[DeviceID]PhysicalVolume, len(lv.id2pv))
for k, v := range lv.id2pv {
dup[k] = v
}
return dup
}
func (lv *LogicalVolume[PhysicalVolume]) ClearMappings() {
lv.logical2physical = nil
lv.physical2logical = nil
}
type Mapping struct {
LAddr LogicalAddr
PAddr QualifiedPhysicalAddr
Size AddrDelta
SizeLocked bool `json:",omitempty"`
Flags containers.Optional[BlockGroupFlags] `json:",omitempty"`
}
func (lv *LogicalVolume[PhysicalVolume]) CouldAddMapping(m Mapping) bool {
return lv.addMapping(m, true) == nil
}
func (lv *LogicalVolume[PhysicalVolume]) AddMapping(m Mapping) error {
return lv.addMapping(m, false)
}
func (lv *LogicalVolume[PhysicalVolume]) addMapping(m Mapping, dryRun bool) error {
lv.init()
// sanity check
if _, haveDev := lv.id2pv[m.PAddr.Dev]; !haveDev {
return fmt.Errorf("(%p).AddMapping: do not have a physical volume with id=%v",
lv, m.PAddr.Dev)
}
// logical2physical
newChunk := chunkMapping{
LAddr: m.LAddr,
PAddrs: []QualifiedPhysicalAddr{m.PAddr},
Size: m.Size,
SizeLocked: m.SizeLocked,
Flags: m.Flags,
}
var logicalOverlaps []chunkMapping
numOverlappingStripes := 0
lv.logical2physical.Subrange(newChunk.compareRange, func(node *containers.RBNode[chunkMapping]) bool {
logicalOverlaps = append(logicalOverlaps, node.Value)
numOverlappingStripes += len(node.Value.PAddrs)
return true
})
var err error
newChunk, err = newChunk.union(logicalOverlaps...)
if err != nil {
return fmt.Errorf("(%p).AddMapping: %w", lv, err)
}
// physical2logical
newExt := devextMapping{
PAddr: m.PAddr.Addr,
LAddr: m.LAddr,
Size: m.Size,
SizeLocked: m.SizeLocked,
Flags: m.Flags,
}
var physicalOverlaps []devextMapping
lv.physical2logical[m.PAddr.Dev].Subrange(newExt.compareRange, func(node *containers.RBNode[devextMapping]) bool {
physicalOverlaps = append(physicalOverlaps, node.Value)
return true
})
newExt, err = newExt.union(physicalOverlaps...)
if err != nil {
return fmt.Errorf("(%p).AddMapping: %w", lv, err)
}
if newChunk.Flags != newExt.Flags {
// If these don't match up, it's a bug in this code.
panic(fmt.Errorf("should not happen: newChunk.Flags:%+v != newExt.Flags:%+v",
newChunk.Flags, newExt.Flags))
}
switch {
case len(physicalOverlaps) == numOverlappingStripes:
// normal case
case len(physicalOverlaps) < numOverlappingStripes:
// .Flags = DUP or RAID{X}
if newChunk.Flags.OK && newChunk.Flags.Val&BLOCK_GROUP_RAID_MASK == 0 {
return fmt.Errorf("multiple stripes but flags=%v does not allow multiple stripes",
newChunk.Flags.Val)
}
case len(physicalOverlaps) > numOverlappingStripes:
// This should not happen because calling .AddMapping
// should update the two in lockstep; if these don't
// match up, it's a bug in this code.
panic(fmt.Errorf("should not happen: len(physicalOverlaps):%d != numOverlappingStripes:%d",
len(physicalOverlaps), numOverlappingStripes))
}
if dryRun {
return nil
}
// optimize
if len(logicalOverlaps) == 1 && reflect.DeepEqual(newChunk, logicalOverlaps[0]) &&
len(physicalOverlaps) == 1 && reflect.DeepEqual(newExt, physicalOverlaps[0]) {
return nil
}
// logical2physical
for _, chunk := range logicalOverlaps {
lv.logical2physical.Delete(lv.logical2physical.Search(chunk.Compare))
}
lv.logical2physical.Insert(newChunk)
// physical2logical
for _, ext := range physicalOverlaps {
lv.physical2logical[m.PAddr.Dev].Delete(lv.physical2logical[m.PAddr.Dev].Search(ext.Compare))
}
lv.physical2logical[m.PAddr.Dev].Insert(newExt)
// sanity check
//
// This is in-theory unnescessary, but that assumes that I
// made no mistakes in my algorithm above.
if os.Getenv("PARANOID") != "" {
if err := lv.fsck(); err != nil {
return err
}
}
// done
return nil
}
func (lv *LogicalVolume[PhysicalVolume]) fsck() error {
physical2logical := make(map[DeviceID]*containers.RBTree[devextMapping])
var err error
lv.logical2physical.Range(func(node *containers.RBNode[chunkMapping]) bool {
chunk := node.Value
for _, stripe := range chunk.PAddrs {
if _, devOK := lv.id2pv[stripe.Dev]; !devOK {
err = fmt.Errorf("(%p).fsck: chunk references physical volume %v which does not exist",
lv, stripe.Dev)
return false
}
if _, exists := physical2logical[stripe.Dev]; !exists {
physical2logical[stripe.Dev] = new(containers.RBTree[devextMapping])
}
physical2logical[stripe.Dev].Insert(devextMapping{
PAddr: stripe.Addr,
LAddr: chunk.LAddr,
Size: chunk.Size,
Flags: chunk.Flags,
})
}
return true
})
if err != nil {
return err
}
if len(lv.physical2logical) != len(physical2logical) {
return fmt.Errorf("(%p).fsck: skew between chunk tree and devext tree",
lv)
}
for devid := range lv.physical2logical {
if !lv.physical2logical[devid].Equal(physical2logical[devid]) {
return fmt.Errorf("(%p).fsck: skew between chunk tree and devext tree",
lv)
}
}
return nil
}
func (lv *LogicalVolume[PhysicalVolume]) Mappings() []Mapping {
var ret []Mapping
lv.logical2physical.Range(func(node *containers.RBNode[chunkMapping]) bool {
chunk := node.Value
for _, slice := range chunk.PAddrs {
ret = append(ret, Mapping{
LAddr: chunk.LAddr,
PAddr: slice,
Size: chunk.Size,
Flags: chunk.Flags,
})
}
return true
})
return ret
}
func (lv *LogicalVolume[PhysicalVolume]) Resolve(laddr LogicalAddr) (paddrs containers.Set[QualifiedPhysicalAddr], maxlen AddrDelta) {
node := lv.logical2physical.Search(func(chunk chunkMapping) int {
return chunkMapping{LAddr: laddr, Size: 1}.compareRange(chunk)
})
if node == nil {
return nil, 0
}
chunk := node.Value
offsetWithinChunk := laddr.Sub(chunk.LAddr)
paddrs = make(containers.Set[QualifiedPhysicalAddr])
maxlen = chunk.Size - offsetWithinChunk
for _, stripe := range chunk.PAddrs {
paddrs.Insert(stripe.Add(offsetWithinChunk))
}
return paddrs, maxlen
}
func (lv *LogicalVolume[PhysicalVolume]) ResolveAny(laddr LogicalAddr, size AddrDelta) (LogicalAddr, QualifiedPhysicalAddr) {
node := lv.logical2physical.Search(func(chunk chunkMapping) int {
return chunkMapping{LAddr: laddr, Size: size}.compareRange(chunk)
})
if node == nil {
return -1, QualifiedPhysicalAddr{0, -1}
}
return node.Value.LAddr, node.Value.PAddrs[0]
}
func (lv *LogicalVolume[PhysicalVolume]) UnResolve(paddr QualifiedPhysicalAddr) LogicalAddr {
node := lv.physical2logical[paddr.Dev].Search(func(ext devextMapping) int {
return devextMapping{PAddr: paddr.Addr, Size: 1}.compareRange(ext)
})
if node == nil {
return -1
}
ext := node.Value
offsetWithinExt := paddr.Addr.Sub(ext.PAddr)
return ext.LAddr.Add(offsetWithinExt)
}
func (lv *LogicalVolume[PhysicalVolume]) ReadAt(dat []byte, laddr LogicalAddr) (int, error) {
done := 0
for done < len(dat) {
n, err := lv.maybeShortReadAt(dat[done:], laddr+LogicalAddr(done))
done += n
if err != nil {
return done, err
}
}
return done, nil
}
var ErrCouldNotMap = errors.New("could not map logical address")
func (lv *LogicalVolume[PhysicalVolume]) maybeShortReadAt(dat []byte, laddr LogicalAddr) (int, error) {
paddrs, maxlen := lv.Resolve(laddr)
if len(paddrs) == 0 {
return 0, fmt.Errorf("read: %w %v", ErrCouldNotMap, laddr)
}
if AddrDelta(len(dat)) > maxlen {
dat = dat[:maxlen]
}
buf := dat
first := true
for paddr := range paddrs {
dev, ok := lv.id2pv[paddr.Dev]
if !ok {
return 0, fmt.Errorf("device=%v does not exist", paddr.Dev)
}
if !first {
buf = make([]byte, len(buf))
}
if _, err := dev.ReadAt(buf, paddr.Addr); err != nil {
return 0, fmt.Errorf("read device=%v paddr=%v: %w", paddr.Dev, paddr.Addr, err)
}
if !first && !bytes.Equal(dat, buf) {
return 0, fmt.Errorf("inconsistent stripes at laddr=%v len=%v", laddr, len(dat))
}
first = false
}
return len(dat), nil
}
func (lv *LogicalVolume[PhysicalVolume]) WriteAt(dat []byte, laddr LogicalAddr) (int, error) {
done := 0
for done < len(dat) {
n, err := lv.maybeShortWriteAt(dat[done:], laddr+LogicalAddr(done))
done += n
if err != nil {
return done, err
}
}
return done, nil
}
func (lv *LogicalVolume[PhysicalVolume]) maybeShortWriteAt(dat []byte, laddr LogicalAddr) (int, error) {
paddrs, maxlen := lv.Resolve(laddr)
if len(paddrs) == 0 {
return 0, fmt.Errorf("write: %w %v", ErrCouldNotMap, laddr)
}
if AddrDelta(len(dat)) > maxlen {
dat = dat[:maxlen]
}
for paddr := range paddrs {
dev, ok := lv.id2pv[paddr.Dev]
if !ok {
return 0, fmt.Errorf("device=%v does not exist", paddr.Dev)
}
if _, err := dev.WriteAt(dat, paddr.Addr); err != nil {
return 0, fmt.Errorf("write device=%v paddr=%v: %w", paddr.Dev, paddr.Addr, err)
}
}
return len(dat), nil
}
|