remove symlinks and fix hardlink unit test

pkg/util/fs_util.go

@@ -60,7 +60,6 @@ func GetFSFromImage(root string, img v1.Image) error {
 
 	fs := map[string]struct{}{}
 	whiteouts := map[string]struct{}{}
-	symlinks := map[string]struct{}{}
 	hardlinks := map[string]*hardlink{}
 
 	for i := len(layers) - 1; i >= 0; i-- {
@@ -106,7 +105,6 @@ func GetFSFromImage(root string, img v1.Image) error {
 					logrus.Debugf("skipping symlink from %s to %s because %s is whitelisted", hdr.Linkname, path, hdr.Linkname)
 					continue
 				}
-				symlinks[filepath.Clean(filepath.Join("/", hdr.Name))] = struct{}{}
 			}
 			if hdr.Typeflag == tar.TypeLink {
 				// If linkname no longer exists, extract hardlink as regular file
@@ -116,7 +114,7 @@ func GetFSFromImage(root string, img v1.Image) error {
 					continue
 				}
 				_, previouslyAdded := fs[linkname]
-				if previouslyAdded || checkSymlinks(linkname, symlinks) || checkWhiteouts(linkname, whiteouts) {
+				if previouslyAdded || checkWhiteouts(linkname, whiteouts) {
 					if h, ok := hardlinks[linkname]; ok {
 						h.links = append(h.links, hdr)
 						continue
@@ -287,15 +285,6 @@ func checkWhiteouts(path string, whiteouts map[string]struct{}) bool {
 	return false
 }
 
-func checkSymlinks(path string, symlinks map[string]struct{}) bool {
-	for sym := range symlinks {
-		if path == sym {
-			return true
-		}
-	}
-	return false
-}
-
 func CheckWhitelist(path string) bool {
 	for _, wl := range whitelist {
 		if HasFilepathPrefix(path, wl) {

pkg/util/fs_util_test.go

@@ -339,7 +339,7 @@ func fileExists(p string) checker {
 	return func(root string, t *testing.T) {
 		_, err := os.Stat(filepath.Join(root, p))
 		if err != nil {
-			t.Fatalf("File does not exist")
+			t.Fatalf("File %s does not exist", filepath.Join(root, p))
 		}
 	}
 }
@@ -381,6 +381,24 @@ func linkPointsTo(src, dst string) checker {
 	}
 }
 
+func filesAreHardlinks(first, second string) checker {
+	return func(root string, t *testing.T) {
+		fi1, err := os.Stat(filepath.Join(root, first))
+		if err != nil {
+			t.Fatalf("error getting file %s", first)
+		}
+		fi2, err := os.Stat(filepath.Join(second))
+		if err != nil {
+			t.Fatalf("error getting file %s", second)
+		}
+		stat1 := getSyscallStat_t(fi1)
+		stat2 := getSyscallStat_t(fi2)
+		if stat1.Ino != stat2.Ino {
+			t.Errorf("%s and %s aren't hardlinks as they dont' have the same inode", first, second)
+		}
+	}
+}
+
 func fileHeader(name string, contents string, mode int64) *tar.Header {
 	return &tar.Header{
 		Name:     name,
@@ -425,6 +443,7 @@ func TestExtractFile(t *testing.T) {
 	type tc struct {
 		name     string
 		hdrs     []*tar.Header
+		tmpdir   string
 		contents []byte
 		checkers []checker
 	}
@@ -496,13 +515,15 @@ func TestExtractFile(t *testing.T) {
 			},
 		},
 		{
-			name: "hardlink",
+			name:   "hardlink",
+			tmpdir: "/tmp/hardlink",
 			hdrs: []*tar.Header{
 				fileHeader("/bin/gzip", "gzip-binary", 0751),
-				hardlinkHeader("/bin/uncompress", "/bin/gzip"),
+				hardlinkHeader("/bin/uncompress", "/tmp/hardlink/bin/gzip"),
 			},
 			checkers: []checker{
-				linkPointsTo("/bin/uncompress", "/bin/gzip"),
+				fileExists("/bin/gzip"),
+				filesAreHardlinks("/bin/uncompress", "/tmp/hardlink/bin/gzip"),
 			},
 		},
 	}
@@ -511,11 +532,19 @@ func TestExtractFile(t *testing.T) {
 		t.Run(tc.name, func(t *testing.T) {
 			tc := tc
 			t.Parallel()
-			r, err := ioutil.TempDir("", "")
-			if err != nil {
-				t.Fatal(err)
+			r := ""
+			var err error
+
+			if tc.tmpdir != "" {
+				r = tc.tmpdir
+			} else {
+				r, err = ioutil.TempDir("", "")
+				if err != nil {
+					t.Fatal(err)
+				}
 			}
 			defer os.RemoveAll(r)
+
 			for _, hdr := range tc.hdrs {
 				if err := extractFile(r, hdr, bytes.NewReader(tc.contents)); err != nil {
 					t.Fatal(err)

pkg/util/tar_util.go

@@ -89,24 +89,32 @@ func Whiteout(p string, w *tar.Writer) error {
 func checkHardlink(p string, hardlinks map[uint64]string, i os.FileInfo) (bool, string) {
 	hardlink := false
 	linkDst := ""
-	if sys := i.Sys(); sys != nil {
-		if stat, ok := sys.(*syscall.Stat_t); ok {
-			nlinks := stat.Nlink
-			if nlinks > 1 {
-				inode := stat.Ino
-				if original, exists := hardlinks[inode]; exists && original != p {
-					hardlink = true
-					logrus.Debugf("%s inode exists in hardlinks map, linking to %s", p, original)
-					linkDst = original
-				} else {
-					hardlinks[inode] = p
-				}
+	stat := getSyscallStat_t(i)
+	if stat != nil {
+		nlinks := stat.Nlink
+		if nlinks > 1 {
+			inode := stat.Ino
+			if original, exists := hardlinks[inode]; exists && original != p {
+				hardlink = true
+				logrus.Debugf("%s inode exists in hardlinks map, linking to %s", p, original)
+				linkDst = original
+			} else {
+				hardlinks[inode] = p
 			}
 		}
 	}
 	return hardlink, linkDst
 }
 
+func getSyscallStat_t(i os.FileInfo) *syscall.Stat_t {
+	if sys := i.Sys(); sys != nil {
+		if stat, ok := sys.(*syscall.Stat_t); ok {
+			return stat
+		}
+	}
+	return nil
+}
+
 // UnpackLocalTarArchive unpacks the tar archive at path to the directory dest
 // Returns true if the path was actually unpacked
 func UnpackLocalTarArchive(path, dest string) error {