Add contrib/findholes

Utility and libs to find hole/data segments using lseek.

4 files changed, 144 insertions, 2 deletions

diff --git a/build.bash b/build.bash
index 57af9eb..427edfd 100755
--- a/build.bash
+++ b/build.bash
@@ -94,8 +94,9 @@ fi
 # Actual "go build" call for gocryptfs
 go build "-ldflags=$GO_LDFLAGS" "$@"
 # Additional binaries
-(cd gocryptfs-xray; go build "-ldflags=$GO_LDFLAGS" "$@")
-(cd contrib/statfs; go build "-ldflags=$GO_LDFLAGS" "$@")
+for d in gocryptfs-xray contrib/statfs contrib/findholes ; do
+	(cd "$d"; go build "-ldflags=$GO_LDFLAGS" "$@")
+done
 
 ./gocryptfs -version
 
diff --git a/contrib/findholes/.gitignore b/contrib/findholes/.gitignore
new file mode 100644
index 0000000..15c37f6
--- /dev/null
+++ b/contrib/findholes/.gitignore
@@ -0,0 +1 @@
+/findholes
diff --git a/contrib/findholes/holes/holes.go b/contrib/findholes/holes/holes.go
new file mode 100644
index 0000000..a2c16de
--- /dev/null
+++ b/contrib/findholes/holes/holes.go
@@ -0,0 +1,106 @@
+// Package holes finds and pretty-prints holes & data sections of a file.
+// Used by TestFileHoleCopy in the gocryptfs test suite.
+package holes
+
+import (
+	"fmt"
+	"syscall"
+)
+
+const (
+	SEEK_DATA = 3
+	SEEK_HOLE = 4
+
+	SegmentHole = SegmentType(100)
+	SegmentData = SegmentType(101)
+	SegmentEOF  = SegmentType(102)
+)
+
+type Segment struct {
+	Offset int64
+	Type   SegmentType
+}
+
+func (s Segment) String() string {
+	return fmt.Sprintf("%10d %v", s.Offset, s.Type)
+}
+
+type SegmentType int
+
+func (s SegmentType) String() string {
+	switch s {
+	case SegmentHole:
+		return "hole"
+	case SegmentData:
+		return "data"
+	case SegmentEOF:
+		return "eof"
+	default:
+		return "???"
+	}
+}
+
+// PrettyPrint pretty-prints the Segments.
+func PrettyPrint(segments []Segment) (out string) {
+	for _, s := range segments {
+		out += "\n" + s.String()
+	}
+	return out
+}
+
+// Find examines the file passed via file descriptor and returns the found holes
+// and data sections.
+func Find(fd int) (segments []Segment, err error) {
+	var st syscall.Stat_t
+	err = syscall.Fstat(fd, &st)
+	if err != nil {
+		return nil, err
+	}
+	totalSize := st.Size
+
+	var cursor int64
+
+	// find out if file starts with data or hole
+	off, err := syscall.Seek(fd, 0, SEEK_DATA)
+	if err == syscall.ENXIO {
+		segments = append(segments,
+			Segment{0, SegmentHole},
+			Segment{totalSize, SegmentEOF})
+		return segments, nil
+	}
+	if err != nil {
+		return nil, err
+	}
+	if off == cursor {
+		segments = append(segments, Segment{0, SegmentData})
+	} else {
+		segments = append(segments,
+			Segment{0, SegmentHole},
+			Segment{totalSize, SegmentData})
+		cursor = off
+	}
+
+	// now we are at the start of data.
+	// find next hole, then next data, then next hole, then next data...
+	for {
+		cursor, err = syscall.Seek(fd, cursor, SEEK_HOLE)
+		if err != nil {
+			return nil, err
+		}
+		if cursor == totalSize {
+			segments = append(segments, Segment{cursor, SegmentEOF})
+			break
+		}
+		segments = append(segments, Segment{cursor, SegmentHole})
+		cursor, err = syscall.Seek(fd, cursor, SEEK_DATA)
+		if err != nil {
+			return nil, err
+		}
+		if cursor == totalSize {
+			segments = append(segments, Segment{cursor, SegmentEOF})
+			break
+		}
+		segments = append(segments, Segment{cursor, SegmentData})
+	}
+	return segments, nil
+}
diff --git a/contrib/findholes/main.go b/contrib/findholes/main.go
new file mode 100644
index 0000000..6f1ae64
--- /dev/null
+++ b/contrib/findholes/main.go
@@ -0,0 +1,34 @@
+// Find and pretty-print holes & data sections of a file.
+package main
+
+import (
+	"flag"
+	"fmt"
+	"os"
+
+	"github.com/rfjakob/gocryptfs/contrib/findholes/holes"
+)
+
+func main() {
+	flag.Parse()
+	if flag.NArg() != 1 {
+		fmt.Printf("Usage: findholes FILE\n")
+		os.Exit(1)
+	}
+
+	f, err := os.Open(flag.Arg(0))
+	if err != nil {
+		// os.Open() gives nicer error messages than syscall.Open()
+		fmt.Println(err)
+		os.Exit(1)
+	}
+	defer f.Close()
+
+	segments, err := holes.Find(int(f.Fd()))
+	if err != nil {
+		fmt.Println(err)
+		os.Exit(1)
+	}
+
+	fmt.Println(holes.PrettyPrint(segments))
+}