geodata: reject entry lengths exceeding asset size

common/geodata/geodat_loader.go

@@ -17,12 +17,12 @@ import (
 )
 
 func checkFile(file, code string) error {
-	r, err := filesystem.OpenAsset(file)
+	r, size, err := filesystem.OpenAsset(file)
 	if err != nil {
 		return errors.New("failed to open ", file).Base(err)
 	}
 	defer r.Close()
-	if _, err := find(r, []byte(code), false); err != nil {
+	if _, err := find(r, []byte(code), false, size); err != nil {
 		return errors.New("failed to check code ", code, " from ", file).Base(err)
 	}
 	return nil
@@ -30,12 +30,12 @@ func checkFile(file, code string) error {
 
 func loadFile(file, code string) ([]byte, error) {
 	runtime.GC() // peak mem
-	r, err := filesystem.OpenAsset(file)
+	r, size, err := filesystem.OpenAsset(file)
 	if err != nil {
 		return nil, errors.New("failed to open ", file).Base(err)
 	}
 	defer r.Close()
-	bs, err := find(r, []byte(code), true)
+	bs, err := find(r, []byte(code), true, size)
 	if err != nil {
 		return nil, errors.New("failed to load code ", code, " from ", file).Base(err)
 	}
@@ -60,13 +60,13 @@ func loadIP(file, code string) ([]*CIDR, error) {
 // unmarshalling it into a []*Domain, so value is only valid during fn.
 func loadSite(file, code, attrs string, fn func(Domain_Type, []byte)) error {
 	runtime.GC() // peak mem
-	r, err := filesystem.OpenAsset(file)
+	r, size, err := filesystem.OpenAsset(file)
 	if err != nil {
 		return errors.New("failed to open ", file).Base(err)
 	}
 	defer r.Close()
 	br := bufio.NewReaderSize(r, 64*1024)
-	n, err := seek(br, []byte(code))
+	n, err := seek(br, []byte(code), size)
 	if err != nil {
 		return errors.New("failed to load code ", code, " from ", file).Base(err)
 	}
@@ -126,9 +126,9 @@ func decodeVarint(br *bufio.Reader) (uint64, error) {
 	return 0, errors.New("varint overflow")
 }
 
-func find(r io.Reader, code []byte, readBody bool) ([]byte, error) {
+func find(r io.Reader, code []byte, readBody bool, size int64) ([]byte, error) {
 	br := bufio.NewReaderSize(r, 64*1024)
-	bodyL, err := seek(br, code)
+	bodyL, err := seek(br, code, size)
 	if err != nil || !readBody {
 		return nil, err
 	}
@@ -140,7 +140,7 @@ func find(r io.Reader, code []byte, readBody bool) ([]byte, error) {
 }
 
 // seek advances br to the body of the entry for code and returns the body length.
-func seek(br *bufio.Reader, code []byte) (int, error) {
+func seek(br *bufio.Reader, code []byte, size int64) (int, error) {
 	codeL := len(code)
 	if codeL == 0 {
 		return 0, errors.New("empty code")
@@ -156,10 +156,10 @@ func seek(br *bufio.Reader, code []byte) (int, error) {
 		if err != nil {
 			return 0, err
 		}
-		bodyL := int(x)
-		if bodyL <= 0 {
-			return 0, errors.New("invalid body length: ", bodyL)
+		if x == 0 || x > uint64(size) || x > uint64(^uint(0)>>1) {
+			return 0, errors.New("invalid body length: ", x)
 		}
+		bodyL := int(x)
 
 		// Peek no more than the buffer holds: a code longer than the buffer cannot match a single
 		// length byte anyway, so a short peek only skips it, as base find (io.ReadFull) does.

common/geodata/geodat_loader_test.go

@@ -28,6 +28,29 @@ func TestLoadFileRejectsVarintOverflow(t *testing.T) {
 	}
 }
 
+func TestLoadFileRejectsOversizedEntry(t *testing.T) {
+	payload := []byte{
+		0x0a, 0x03, 0x0a, 0x01, 'A', // A valid entry before the oversized one.
+		0x0a, 0x80, 0x80, 0x80, 0x80, 0x04, // Entry length: 1 GiB.
+		0x0a, 0x04, 'T', 'E', 'S', 'T',
+	}
+	dir := t.TempDir()
+	t.Setenv("xray.location.asset", dir)
+	if err := os.WriteFile(filepath.Join(dir, "oversized.dat"), payload, 0o600); err != nil {
+		t.Fatal(err)
+	}
+	// Check without allocating the body first, so a regression fails safely.
+	if err := checkFile("oversized.dat", "TEST"); err == nil {
+		t.Fatal("accepted an entry length larger than the file")
+	}
+	if _, err := loadFile("oversized.dat", "TEST"); err == nil || !strings.Contains(err.Error(), "invalid body length") {
+		t.Fatalf("want invalid body length, got %v", err)
+	}
+	if err := loadSite("oversized.dat", "TEST", "", func(Domain_Type, []byte) {}); err == nil || !strings.Contains(err.Error(), "invalid body length") {
+		t.Fatalf("want invalid body length, got %v", err)
+	}
+}
+
 type siteEntry struct {
 	Type  Domain_Type
 	Value string