file reading!

4 files changed, 153 insertions, 43 deletions

diff --git a/cmd/btrfs-mount/subvol.go b/cmd/btrfs-mount/subvol.go
index 4e741ad..1d5e9db 100644
--- a/cmd/btrfs-mount/subvol.go
+++ b/cmd/btrfs-mount/subvol.go
@@ -3,7 +3,9 @@ package main
 import (
 	"context"
 	"fmt"
+	"io"
 	"reflect"
+	"sort"
 	"sync"
 
 	"github.com/datawire/dlib/dcontext"
@@ -35,9 +37,15 @@ type dir struct {
 	ChildrenByIndex map[uint64]btrfsitem.DirEntry
 }
 
+type fileExtent struct {
+	OffsetWithinFile int64
+	btrfsitem.FileExtent
+}
+
 type file struct {
 	fullInode
-	// TODO
+	Extents []fileExtent
+	FS      *btrfs.FS
 }
 
 type Subvolume struct {
@@ -296,7 +304,8 @@ func (sv *Subvolume) loadFile(inode btrfs.ObjID) (*file, error) {
 			return
 		}
 		val.fullInode = *fullInode
-		// TODO
+		val.FS = sv.FS
+		val.populate()
 		return
 	})
 	if val.Inode == 0 {
@@ -304,3 +313,100 @@ func (sv *Subvolume) loadFile(inode btrfs.ObjID) (*file, error) {
 	}
 	return val, nil
 }
+
+func (ret *file) populate() {
+	for _, item := range ret.OtherItems {
+		switch item.Head.Key.ItemType {
+		case btrfsitem.INODE_REF_KEY:
+			// TODO
+		case btrfsitem.EXTENT_DATA_KEY:
+			ret.Extents = append(ret.Extents, fileExtent{
+				OffsetWithinFile: int64(item.Head.Key.Offset),
+				FileExtent:       item.Body.(btrfsitem.FileExtent),
+			})
+		default:
+			panic(fmt.Errorf("TODO: handle item type %v", item.Head.Key.ItemType))
+		}
+	}
+
+	// These should already be sorted, because of the nature of
+	// the btree; but this is a recovery tool for corrupt
+	// filesystems, so go ahead and ensure that it's sorted.
+	sort.Slice(ret.Extents, func(i, j int) bool {
+		return ret.Extents[i].OffsetWithinFile < ret.Extents[j].OffsetWithinFile
+	})
+
+	pos := int64(0)
+	for _, extent := range ret.Extents {
+		if extent.OffsetWithinFile != pos {
+			if extent.OffsetWithinFile > pos {
+				ret.Errs = append(ret.Errs, fmt.Errorf("extent gap from %v to %v",
+					pos, extent.OffsetWithinFile))
+			} else {
+				ret.Errs = append(ret.Errs, fmt.Errorf("extent overlap from %v to %v",
+					extent.OffsetWithinFile, pos))
+			}
+		}
+		size, err := extent.Size()
+		if err != nil {
+			ret.Errs = append(ret.Errs, fmt.Errorf("extent %v: %w", extent.OffsetWithinFile, err))
+		}
+		pos += size
+	}
+	if ret.InodeItem != nil && pos != ret.InodeItem.Size {
+		if ret.InodeItem.Size > pos {
+			ret.Errs = append(ret.Errs, fmt.Errorf("extent gap from %v to %v",
+				pos, ret.InodeItem.Size))
+		} else {
+			ret.Errs = append(ret.Errs, fmt.Errorf("extent mapped past end of file from %v to %v",
+				ret.InodeItem.Size, pos))
+		}
+	}
+}
+
+func (file *file) ReadAt(dat []byte, off int64) (int, error) {
+	// These stateles maybe-short-reads each do an O(n) extent
+	// lookup, so reading a file is O(n^2), but we expect n to be
+	// small, so whatev.  Turn file.Extents it in to an rbtree if
+	// it becomes a problem.
+	done := 0
+	for done < len(dat) {
+		n, err := file.maybeShortReadAt(dat[done:], off+int64(done))
+		done += n
+		if err != nil {
+			return done, err
+		}
+	}
+	return done, nil
+}
+
+func (file *file) maybeShortReadAt(dat []byte, off int64) (int, error) {
+	for _, extent := range file.Extents {
+		extBeg := extent.OffsetWithinFile
+		if extBeg > off {
+			break
+		}
+		extLen, err := extent.Size()
+		if err != nil {
+			continue
+		}
+		extEnd := extBeg + extLen
+		if extEnd <= off {
+			continue
+		}
+		offsetWithinExt := off - extent.OffsetWithinFile
+		readSize := util.Min(int64(len(dat)), extLen-offsetWithinExt)
+		switch extent.Type {
+		case btrfsitem.FILE_EXTENT_INLINE:
+			return copy(dat, extent.BodyInline[offsetWithinExt:offsetWithinExt+readSize]), nil
+		case btrfsitem.FILE_EXTENT_REG, btrfsitem.FILE_EXTENT_PREALLOC:
+			return file.FS.ReadAt(dat[:readSize],
+				extent.BodyExtent.DiskByteNr.
+					Add(extent.BodyExtent.Offset).
+					Add(btrfsvol.AddrDelta(offsetWithinExt)))
+		}
+	}
+	return 0, fmt.Errorf("read: could not map position %v", off)
+}
+
+var _ io.ReaderAt = (*file)(nil)
diff --git a/cmd/btrfs-mount/subvol_fuse.go b/cmd/btrfs-mount/subvol_fuse.go
index 7500e0b..6f3f267 100644
--- a/cmd/btrfs-mount/subvol_fuse.go
+++ b/cmd/btrfs-mount/subvol_fuse.go
@@ -197,11 +197,25 @@ func (sv *Subvolume) OpenFile(_ context.Context, op *fuseops.OpenFileOp) error {
 	return nil
 }
 func (sv *Subvolume) ReadFile(_ context.Context, op *fuseops.ReadFileOp) error {
-	_, ok := sv.fileHandles.Load(op.Handle)
+	state, ok := sv.fileHandles.Load(op.Handle)
 	if !ok {
 		return syscall.EBADF
 	}
-	return syscall.ENOSYS
+
+	size := op.Size
+	var dat []byte
+	if op.Dst != nil {
+		size = util.Min(int64(len(op.Dst)), size)
+		dat = op.Dst
+	} else {
+		dat = make([]byte, op.Size)
+		op.Data = [][]byte{dat}
+	}
+
+	var err error
+	op.BytesRead, err = state.File.ReadAt(dat, op.Offset)
+
+	return err
 }
 func (sv *Subvolume) ReleaseFileHandle(_ context.Context, op *fuseops.ReleaseFileHandleOp) error {
 	_, ok := sv.fileHandles.LoadAndDelete(op.Handle)
diff --git a/pkg/btrfs/btrfsitem/item_fileextent.go b/pkg/btrfs/btrfsitem/item_fileextent.go
index f3cbffe..f480207 100644
--- a/pkg/btrfs/btrfsitem/item_fileextent.go
+++ b/pkg/btrfs/btrfsitem/item_fileextent.go
@@ -24,26 +24,17 @@ type FileExtent struct { // EXTENT_DATA=108
 	binstruct.End `bin:"off=0x15"`
 
 	// only one of these, depending on .Type
-	BodyInline []byte `bin:"-"`
-	BodyReg    struct {
-		// Position of extent within the device
+	BodyInline []byte   `bin:"-"` // .Type == FILE_EXTENT_INLINE
+	BodyExtent struct { // .Type == FILE_EXTENT_REG or FILE_EXTENT_PREALLOC
+		// Position and size of extent within the device
 		DiskByteNr   btrfsvol.LogicalAddr `bin:"off=0x0, siz=0x8"`
 		DiskNumBytes btrfsvol.AddrDelta   `bin:"off=0x8, siz=0x8"`
 
 		// Position of data within the extent
-		Offset   btrfsvol.AddrDelta `bin:"off=0x10, siz=0x8"`
-		NumBytes btrfsvol.AddrDelta `bin:"off=0x18, siz=0x8"`
+		Offset btrfsvol.AddrDelta `bin:"off=0x10, siz=0x8"`
 
-		binstruct.End `bin:"off=0x20"`
-	} `bin:"-"`
-	BodyPrealloc struct {
-		// Position of extent within the device
-		DiskByteNr   btrfsvol.LogicalAddr `bin:"off=0x0, siz=0x8"`
-		DiskNumBytes btrfsvol.AddrDelta   `bin:"off=0x8, siz=0x8"`
-
-		// Position of data within the extent
-		Offset   btrfsvol.AddrDelta `bin:"off=0x10, siz=0x8"`
-		NumBytes btrfsvol.AddrDelta `bin:"off=0x18, siz=0x8"`
+		// Decompressed/unencrypted size
+		NumBytes int64 `bin:"off=0x18, siz=0x8"`
 
 		binstruct.End `bin:"off=0x20"`
 	} `bin:"-"`
@@ -58,14 +49,8 @@ func (o *FileExtent) UnmarshalBinary(dat []byte) (int, error) {
 	case FILE_EXTENT_INLINE:
 		o.BodyInline = dat[n:]
 		n += len(o.BodyInline)
-	case FILE_EXTENT_REG:
-		_n, err := binstruct.Unmarshal(dat[n:], &o.BodyReg)
-		n += _n
-		if err != nil {
-			return n, err
-		}
-	case FILE_EXTENT_PREALLOC:
-		_n, err := binstruct.Unmarshal(dat[n:], &o.BodyPrealloc)
+	case FILE_EXTENT_REG, FILE_EXTENT_PREALLOC:
+		_n, err := binstruct.Unmarshal(dat[n:], &o.BodyExtent)
 		n += _n
 		if err != nil {
 			return n, err
@@ -84,14 +69,8 @@ func (o FileExtent) MarshalBinary() ([]byte, error) {
 	switch o.Type {
 	case FILE_EXTENT_INLINE:
 		dat = append(dat, o.BodyInline...)
-	case FILE_EXTENT_REG:
-		bs, err := binstruct.Marshal(o.BodyReg)
-		dat = append(dat, bs...)
-		if err != nil {
-			return dat, err
-		}
-	case FILE_EXTENT_PREALLOC:
-		bs, err := binstruct.Marshal(o.BodyPrealloc)
+	case FILE_EXTENT_REG, FILE_EXTENT_PREALLOC:
+		bs, err := binstruct.Marshal(o.BodyExtent)
 		dat = append(dat, bs...)
 		if err != nil {
 			return dat, err
@@ -110,6 +89,17 @@ const (
 	FILE_EXTENT_PREALLOC
 )
 
+func (o FileExtent) Size() (int64, error) {
+	switch o.Type {
+	case FILE_EXTENT_INLINE:
+		return int64(len(o.BodyInline)), nil
+	case FILE_EXTENT_REG, FILE_EXTENT_PREALLOC:
+		return o.BodyExtent.NumBytes, nil
+	default:
+		return 0, fmt.Errorf("unknown file extent type %v", o.Type)
+	}
+}
+
 func (fet FileExtentType) String() string {
 	names := map[FileExtentType]string{
 		FILE_EXTENT_INLINE:   "inline",
diff --git a/pkg/btrfsmisc/print_tree.go b/pkg/btrfsmisc/print_tree.go
index 6b4ff96..61c5423 100644
--- a/pkg/btrfsmisc/print_tree.go
+++ b/pkg/btrfsmisc/print_tree.go
@@ -156,18 +156,18 @@ func PrintTree(fs *btrfs.FS, root btrfsvol.LogicalAddr) error {
 						len(body.BodyInline), body.RAMBytes, body.Compression)
 				case btrfsitem.FILE_EXTENT_PREALLOC:
 					fmt.Printf("\t\tprealloc data disk byte %v nr %v\n",
-						body.BodyPrealloc.DiskByteNr,
-						body.BodyPrealloc.DiskNumBytes)
+						body.BodyExtent.DiskByteNr,
+						body.BodyExtent.DiskNumBytes)
 					fmt.Printf("\t\tprealloc data offset %v nr %v\n",
-						body.BodyPrealloc.Offset,
-						body.BodyPrealloc.NumBytes)
+						body.BodyExtent.Offset,
+						body.BodyExtent.NumBytes)
 				case btrfsitem.FILE_EXTENT_REG:
 					fmt.Printf("\t\textent data disk byte %d nr %d\n",
-						body.BodyReg.DiskByteNr,
-						body.BodyReg.DiskNumBytes)
+						body.BodyExtent.DiskByteNr,
+						body.BodyExtent.DiskNumBytes)
 					fmt.Printf("\t\textent data offset %d nr %d ram %v\n",
-						body.BodyReg.Offset,
-						body.BodyReg.NumBytes,
+						body.BodyExtent.Offset,
+						body.BodyExtent.NumBytes,
 						body.RAMBytes)
 					fmt.Printf("\t\textent compression %v\n",
 						body.Compression)