summaryrefslogtreecommitdiff
path: root/lib/btrfsprogs/btrfsinspect/mount.go
blob: 2a0b232fd8c2610e17349de3c490d7fb5b9d1268 (plain)
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
// Copyright (C) 2022  Luke Shumaker <lukeshu@lukeshu.com>
//
// SPDX-License-Identifier: GPL-2.0-or-later

package btrfsinspect

import (
	"context"
	"errors"
	"fmt"
	"io"
	"path/filepath"
	"sync"
	"sync/atomic"
	"syscall"

	"github.com/datawire/dlib/dcontext"
	"github.com/datawire/dlib/dgroup"
	"github.com/datawire/dlib/dlog"
	"github.com/jacobsa/fuse"
	"github.com/jacobsa/fuse/fuseops"
	"github.com/jacobsa/fuse/fuseutil"

	"git.lukeshu.com/btrfs-progs-ng/lib/btrfs"
	"git.lukeshu.com/btrfs-progs-ng/lib/btrfs/btrfsitem"
	"git.lukeshu.com/btrfs-progs-ng/lib/btrfs/btrfsprim"
	"git.lukeshu.com/btrfs-progs-ng/lib/btrfsprogs/btrfsutil"
	"git.lukeshu.com/btrfs-progs-ng/lib/containers"
	"git.lukeshu.com/btrfs-progs-ng/lib/maps"
	"git.lukeshu.com/btrfs-progs-ng/lib/slices"
)

func MountRO(ctx context.Context, fs *btrfs.FS, mountpoint string, noChecksums bool) error {
	pvs := fs.LV.PhysicalVolumes()
	if len(pvs) < 1 {
		return errors.New("no devices")
	}

	deviceName := pvs[maps.SortedKeys(pvs)[0]].Name()
	if abs, err := filepath.Abs(deviceName); err == nil {
		deviceName = abs
	}

	rootSubvol := &subvolume{
		Subvolume: btrfs.Subvolume{
			FS:          btrfsutil.NewBrokenTrees(ctx, fs),
			TreeID:      btrfsprim.FS_TREE_OBJECTID,
			NoChecksums: noChecksums,
		},
		DeviceName: deviceName,
		Mountpoint: mountpoint,
	}
	return rootSubvol.Run(ctx)
}

func fuseMount(ctx context.Context, mountpoint string, server fuse.Server, cfg *fuse.MountConfig) error {
	grp := dgroup.NewGroup(ctx, dgroup.GroupConfig{
		// Allow mountHandle.Join() returning to cause the
		// "unmount" goroutine to quit.
		ShutdownOnNonError: true,
	})
	mounted := uint32(1)
	grp.Go("unmount", func(ctx context.Context) error {
		<-ctx.Done()
		var err error
		var gotNil bool
		// Keep retrying, because the FS might be busy.
		for atomic.LoadUint32(&mounted) != 0 {
			if _err := fuse.Unmount(mountpoint); _err == nil {
				gotNil = true
			} else if !gotNil {
				err = _err
			}
		}
		if gotNil {
			return nil
		}
		return err
	})
	grp.Go("mount", func(ctx context.Context) error {
		defer atomic.StoreUint32(&mounted, 0)

		cfg.OpContext = ctx
		cfg.ErrorLogger = dlog.StdLogger(ctx, dlog.LogLevelError)
		cfg.DebugLogger = dlog.StdLogger(ctx, dlog.LogLevelDebug)

		mountHandle, err := fuse.Mount(mountpoint, server, cfg)
		if err != nil {
			return err
		}
		dlog.Infof(ctx, "mounted %q", mountpoint)
		return mountHandle.Join(dcontext.HardContext(ctx))
	})
	return grp.Wait()
}

type dirState struct {
	Dir *btrfs.Dir
}

type fileState struct {
	File *btrfs.File
}

type subvolume struct {
	btrfs.Subvolume
	DeviceName string
	Mountpoint string

	fuseutil.NotImplementedFileSystem
	lastHandle  uint64
	dirHandles  containers.SyncMap[fuseops.HandleID, *dirState]
	fileHandles containers.SyncMap[fuseops.HandleID, *fileState]

	subvolMu sync.Mutex
	subvols  containers.Set[string]
	grp      *dgroup.Group
}

func (sv *subvolume) Run(ctx context.Context) error {
	sv.grp = dgroup.NewGroup(ctx, dgroup.GroupConfig{})
	sv.grp.Go("self", func(ctx context.Context) error {
		cfg := &fuse.MountConfig{
			FSName:  sv.DeviceName,
			Subtype: "btrfs",

			ReadOnly: true,

			Options: map[string]string{
				"allow_other": "",
			},
		}
		return fuseMount(ctx, sv.Mountpoint, fuseutil.NewFileSystemServer(sv), cfg)
	})
	return sv.grp.Wait()
}

func (sv *subvolume) newHandle() fuseops.HandleID {
	return fuseops.HandleID(atomic.AddUint64(&sv.lastHandle, 1))
}

func inodeItemToFUSE(itemBody btrfsitem.Inode) fuseops.InodeAttributes {
	return fuseops.InodeAttributes{
		Size:  uint64(itemBody.Size),
		Nlink: uint32(itemBody.NLink),
		Mode:  uint32(itemBody.Mode),
		//RDev: itemBody.Rdev, // jacobsa/fuse doesn't expose rdev
		Atime: itemBody.ATime.ToStd(),
		Mtime: itemBody.MTime.ToStd(),
		Ctime: itemBody.CTime.ToStd(),
		//Crtime: itemBody.OTime,
		Uid: uint32(itemBody.UID),
		Gid: uint32(itemBody.GID),
	}
}

func (sv *subvolume) LoadDir(inode btrfsprim.ObjID) (val *btrfs.Dir, err error) {
	val, err = sv.Subvolume.LoadDir(inode)
	if val != nil {
		haveSubvolumes := false
		for _, index := range maps.SortedKeys(val.ChildrenByIndex) {
			entry := val.ChildrenByIndex[index]
			if entry.Location.ItemType == btrfsitem.ROOT_ITEM_KEY {
				haveSubvolumes = true
				break
			}
		}
		if haveSubvolumes {
			abspath, _err := val.AbsPath()
			if _err != nil {
				return
			}
			sv.subvolMu.Lock()
			for _, index := range maps.SortedKeys(val.ChildrenByIndex) {
				entry := val.ChildrenByIndex[index]
				if entry.Location.ItemType != btrfsitem.ROOT_ITEM_KEY {
					continue
				}
				if sv.subvols == nil {
					sv.subvols = make(containers.Set[string])
				}
				subMountpoint := filepath.Join(abspath, string(entry.Name))
				if !sv.subvols.Has(subMountpoint) {
					sv.subvols.Insert(subMountpoint)
					workerName := fmt.Sprintf("%d-%s", val.Inode, filepath.Base(subMountpoint))
					sv.grp.Go(workerName, func(ctx context.Context) error {
						subSv := &subvolume{
							Subvolume: btrfs.Subvolume{
								FS:          sv.FS,
								TreeID:      entry.Location.ObjectID,
								NoChecksums: sv.NoChecksums,
							},
							DeviceName: sv.DeviceName,
							Mountpoint: filepath.Join(sv.Mountpoint, subMountpoint[1:]),
						}
						return subSv.Run(ctx)
					})
				}
			}
			sv.subvolMu.Unlock()
		}
	}
	return
}

func (sv *subvolume) StatFS(_ context.Context, op *fuseops.StatFSOp) error {
	// See linux.git/fs/btrfs/super.c:btrfs_statfs()
	sb, err := sv.FS.Superblock()
	if err != nil {
		return err
	}

	op.IoSize = sb.SectorSize
	op.BlockSize = sb.SectorSize
	op.Blocks = sb.TotalBytes / uint64(sb.SectorSize) // TODO: adjust for RAID type
	//op.BlocksFree = TODO

	// btrfs doesn't have a fixed number of inodes
	op.Inodes = 0
	op.InodesFree = 0

	// jacobsa/fuse doesn't expose namelen, instead hard-coding it
	// to 255.  Which is fine by us, because that's what it is for
	// btrfs.

	return nil
}

func (sv *subvolume) LookUpInode(_ context.Context, op *fuseops.LookUpInodeOp) error {
	if op.Parent == fuseops.RootInodeID {
		parent, err := sv.GetRootInode()
		if err != nil {
			return err
		}
		op.Parent = fuseops.InodeID(parent)
	}

	dir, err := sv.LoadDir(btrfsprim.ObjID(op.Parent))
	if err != nil {
		return err
	}
	entry, ok := dir.ChildrenByName[op.Name]
	if !ok {
		return syscall.ENOENT
	}
	if entry.Location.ItemType != btrfsitem.INODE_ITEM_KEY {
		// Subvolume
		//
		// Because each subvolume has its own pool of inodes
		// (as in 2 different subvolumes can have files with
		// te same inode number), so to represent that to FUSE
		// we need to have this be a full separate mountpoint.
		//
		// I'd want to return EIO or EINTR or something here,
		// but both the FUSE userspace tools and the kernel
		// itself stat the mountpoint before mounting it, so
		// we've got to return something bogus here to let
		// that mount happen.
		op.Entry = fuseops.ChildInodeEntry{
			Child: 2, // an inode number that a real file will never have
			Attributes: fuseops.InodeAttributes{
				Nlink: 1,
				Mode:  uint32(btrfsitem.ModeFmtDir | 0700),
			},
		}
		return nil
	}
	bareInode, err := sv.LoadBareInode(entry.Location.ObjectID)
	if err != nil {
		return err
	}
	op.Entry = fuseops.ChildInodeEntry{
		Child:      fuseops.InodeID(entry.Location.ObjectID),
		Generation: fuseops.GenerationNumber(bareInode.InodeItem.Sequence),
		Attributes: inodeItemToFUSE(*bareInode.InodeItem),
	}
	return nil
}

func (sv *subvolume) GetInodeAttributes(_ context.Context, op *fuseops.GetInodeAttributesOp) error {
	if op.Inode == fuseops.RootInodeID {
		inode, err := sv.GetRootInode()
		if err != nil {
			return err
		}
		op.Inode = fuseops.InodeID(inode)
	}

	bareInode, err := sv.LoadBareInode(btrfsprim.ObjID(op.Inode))
	if err != nil {
		return err
	}

	op.Attributes = inodeItemToFUSE(*bareInode.InodeItem)
	return nil
}

func (sv *subvolume) OpenDir(_ context.Context, op *fuseops.OpenDirOp) error {
	if op.Inode == fuseops.RootInodeID {
		inode, err := sv.GetRootInode()
		if err != nil {
			return err
		}
		op.Inode = fuseops.InodeID(inode)
	}

	dir, err := sv.LoadDir(btrfsprim.ObjID(op.Inode))
	if err != nil {
		return err
	}
	handle := sv.newHandle()
	sv.dirHandles.Store(handle, &dirState{
		Dir: dir,
	})
	op.Handle = handle
	return nil
}
func (sv *subvolume) ReadDir(_ context.Context, op *fuseops.ReadDirOp) error {
	state, ok := sv.dirHandles.Load(op.Handle)
	if !ok {
		return syscall.EBADF
	}
	origOffset := op.Offset
	for _, index := range maps.SortedKeys(state.Dir.ChildrenByIndex) {
		if index < uint64(origOffset) {
			continue
		}
		entry := state.Dir.ChildrenByIndex[index]
		n := fuseutil.WriteDirent(op.Dst[op.BytesRead:], fuseutil.Dirent{
			Offset: fuseops.DirOffset(index + 1),
			Inode:  fuseops.InodeID(entry.Location.ObjectID),
			Name:   string(entry.Name),
			Type: map[btrfsitem.FileType]fuseutil.DirentType{
				btrfsitem.FT_UNKNOWN:  fuseutil.DT_Unknown,
				btrfsitem.FT_REG_FILE: fuseutil.DT_File,
				btrfsitem.FT_DIR:      fuseutil.DT_Directory,
				btrfsitem.FT_CHRDEV:   fuseutil.DT_Char,
				btrfsitem.FT_BLKDEV:   fuseutil.DT_Block,
				btrfsitem.FT_FIFO:     fuseutil.DT_FIFO,
				btrfsitem.FT_SOCK:     fuseutil.DT_Socket,
				btrfsitem.FT_SYMLINK:  fuseutil.DT_Link,
			}[entry.Type],
		})
		if n == 0 {
			break
		}
		op.BytesRead += n
	}
	return nil
}
func (sv *subvolume) ReleaseDirHandle(_ context.Context, op *fuseops.ReleaseDirHandleOp) error {
	_, ok := sv.dirHandles.LoadAndDelete(op.Handle)
	if !ok {
		return syscall.EBADF
	}
	return nil
}

func (sv *subvolume) OpenFile(_ context.Context, op *fuseops.OpenFileOp) error {
	file, err := sv.LoadFile(btrfsprim.ObjID(op.Inode))
	if err != nil {
		return err
	}
	handle := sv.newHandle()
	sv.fileHandles.Store(handle, &fileState{
		File: file,
	})
	op.Handle = handle
	op.KeepPageCache = true
	return nil
}
func (sv *subvolume) ReadFile(_ context.Context, op *fuseops.ReadFileOp) error {
	state, ok := sv.fileHandles.Load(op.Handle)
	if !ok {
		return syscall.EBADF
	}

	var dat []byte
	if op.Dst != nil {
		size := slices.Min(int64(len(op.Dst)), op.Size)
		dat = op.Dst[:size]
	} else {
		dat = make([]byte, op.Size)
		op.Data = [][]byte{dat}
	}

	var err error
	op.BytesRead, err = state.File.ReadAt(dat, op.Offset)
	if errors.Is(err, io.EOF) {
		err = nil
	}

	return err
}
func (sv *subvolume) ReleaseFileHandle(_ context.Context, op *fuseops.ReleaseFileHandleOp) error {
	_, ok := sv.fileHandles.LoadAndDelete(op.Handle)
	if !ok {
		return syscall.EBADF
	}
	return nil
}

func (sv *subvolume) ReadSymlink(_ context.Context, op *fuseops.ReadSymlinkOp) error {
	file, err := sv.LoadFile(btrfsprim.ObjID(op.Inode))
	if err != nil {
		return err
	}
	reader := io.NewSectionReader(file, 0, file.InodeItem.Size)
	tgt, err := io.ReadAll(reader)
	if err != nil {
		return err
	}
	op.Target = string(tgt)
	return nil
}

func (sv *subvolume) ListXattr(_ context.Context, op *fuseops.ListXattrOp) error {
	if op.Inode == fuseops.RootInodeID {
		inode, err := sv.GetRootInode()
		if err != nil {
			return err
		}
		op.Inode = fuseops.InodeID(inode)
	}

	fullInode, err := sv.LoadFullInode(btrfsprim.ObjID(op.Inode))
	if err != nil {
		return err
	}

	size := 0
	for name := range fullInode.XAttrs {
		size += len(name) + 1
	}
	if len(op.Dst) < size {
		return syscall.ERANGE
	}

	op.BytesRead = size
	n := 0
	for _, name := range maps.SortedKeys(fullInode.XAttrs) {
		n += copy(op.Dst[n:], name)
		op.Dst[n] = 0
		n++
	}
	return nil
}

func (sv *subvolume) GetXattr(_ context.Context, op *fuseops.GetXattrOp) error {
	if op.Inode == fuseops.RootInodeID {
		inode, err := sv.GetRootInode()
		if err != nil {
			return err
		}
		op.Inode = fuseops.InodeID(inode)
	}

	fullInode, err := sv.LoadFullInode(btrfsprim.ObjID(op.Inode))
	if err != nil {
		return err
	}

	val, ok := fullInode.XAttrs[op.Name]
	if !ok {
		return syscall.ENODATA
	}

	if len(op.Dst) < len(val) {
		return syscall.ERANGE
	}

	op.BytesRead = len(val)
	copy(op.Dst, val)
	return nil
}

func (sv *subvolume) Destroy() {}