prevent symlink loop recursion in directory traversal

Author: jmestwa-coder

fuzzing/replay/file_util.cc

@@ -21,7 +21,9 @@
 
 #include <cerrno>
 #include <cstdio>
+#include <set>
 #include <string>
+#include <utility>
 
 #include "absl/functional/function_ref.h"
 #include "absl/status/status.h"
@@ -36,7 +38,33 @@ namespace {
 
 absl::Status TraverseDirectory(
     absl::string_view path,
-    absl::FunctionRef<void(absl::string_view, const struct stat&)> callback) {
+    absl::FunctionRef<void(absl::string_view, const struct stat&)> callback,
+    std::set<std::pair<dev_t, ino_t>>& visited);
+
+absl::Status YieldFilesInternal(
+    absl::string_view path,
+    absl::FunctionRef<void(absl::string_view, const struct stat&)> callback,
+    std::set<std::pair<dev_t, ino_t>>& visited) {
+  struct stat path_stat;
+  if (stat(std::string(path).c_str(), &path_stat) < 0) {
+    return ErrnoStatus(absl::StrCat("could not stat ", path), errno);
+  }
+  if (S_ISDIR(path_stat.st_mode)) {
+    auto dir_id = std::make_pair(path_stat.st_dev, path_stat.st_ino);
+    // Prevent infinite recursion by tracking visited directories (dev,inode).
+    if (!visited.insert(dir_id).second) {
+      return absl::OkStatus();
+    }
+    return TraverseDirectory(path, callback, visited);
+  }
+  callback(path, path_stat);
+  return absl::OkStatus();
+}
+
+absl::Status TraverseDirectory(
+    absl::string_view path,
+    absl::FunctionRef<void(absl::string_view, const struct stat&)> callback,
+    std::set<std::pair<dev_t, ino_t>>& visited) {
   DIR* dir = opendir(std::string(path).c_str());
   if (!dir) {
     return ErrnoStatus(absl::StrCat("could not open directory ", path), errno);
@@ -58,7 +86,7 @@ absl::Status TraverseDirectory(
       continue;
     }
     const std::string entry_path = absl::StrCat(path, "/", entry_name);
-    status.Update(YieldFiles(entry_path, callback));
+    status.Update(YieldFilesInternal(entry_path, callback, visited));
   }
   closedir(dir);
   return status;
@@ -69,15 +97,8 @@ absl::Status TraverseDirectory(
 absl::Status YieldFiles(
     absl::string_view path,
     absl::FunctionRef<void(absl::string_view, const struct stat&)> callback) {
-  struct stat path_stat;
-  if (stat(std::string(path).c_str(), &path_stat) < 0) {
-    return ErrnoStatus(absl::StrCat("could not stat ", path), errno);
-  }
-  if (S_ISDIR(path_stat.st_mode)) {
-    return TraverseDirectory(path, callback);
-  }
-  callback(path, path_stat);
-  return absl::OkStatus();
+  std::set<std::pair<dev_t, ino_t>> visited;
+  return YieldFilesInternal(path, callback, visited);
 }
 
 absl::Status SetFileContents(absl::string_view path,

fuzzing/replay/file_util_test.cc

@@ -16,8 +16,11 @@
 
 #include <sys/stat.h>
 #include <sys/types.h>
+#include <unistd.h>
 
+#include <cerrno>
 #include <cstdlib>
+#include <cstring>
 #include <functional>
 #include <string>
 #include <vector>
@@ -113,6 +116,43 @@ TEST(YieldFilesTest, YieldsHiddenFilesAndDirs) {
   EXPECT_THAT(collected_paths, testing::SizeIs(2));
 }
 
+TEST(YieldFilesTest, DoesNotRecurseThroughSymlinkLoop) {
+  const std::string root_dir =
+      absl::StrCat(getenv("TEST_TMPDIR"), "/symlink-loop-root");
+  ASSERT_EQ(mkdir(root_dir.c_str(), 0755), 0);
+  const std::string dir_a = absl::StrCat(root_dir, "/dirA");
+  ASSERT_EQ(mkdir(dir_a.c_str(), 0755), 0);
+  const std::string dir_b = absl::StrCat(root_dir, "/dirB");
+  ASSERT_EQ(mkdir(dir_b.c_str(), 0755), 0);
+
+  // Normal files that must each be yielded exactly once.
+  const std::string file_a = absl::StrCat(root_dir, "/a");
+  const std::string file_b = absl::StrCat(root_dir, "/b");
+  ASSERT_TRUE(SetFileContents(file_a, "foo").ok());
+  ASSERT_TRUE(SetFileContents(file_b, "bar").ok());
+
+  // Build a symlink cycle: dirA/toB -> dirB and dirB/toA -> dirA. Without cycle
+  // detection, traversal would recurse forever (dirA -> dirB -> dirA -> ...).
+  const std::string link_to_b = absl::StrCat(dir_a, "/toB");
+  if (symlink(dir_b.c_str(), link_to_b.c_str()) != 0) {
+    GTEST_SKIP() << "symlink unsupported in this environment: "
+                 << std::strerror(errno);
+  }
+  const std::string link_to_a = absl::StrCat(dir_b, "/toA");
+  ASSERT_EQ(symlink(dir_a.c_str(), link_to_a.c_str()), 0);
+
+  std::vector<std::string> collected_paths;
+  const absl::Status status =
+      YieldFiles(root_dir, CollectPathsCallback(&collected_paths));
+  EXPECT_TRUE(status.ok());
+  // Only the two regular files are yielded; directories (including those reached
+  // through the symlinks) never invoke the callback, and the cycle is visited
+  // at most once, so the result is fully deterministic regardless of readdir
+  // ordering.
+  EXPECT_THAT(collected_paths,
+              testing::UnorderedElementsAre(file_a, file_b));
+}
+
 }  // namespace
 
 }  // namespace fuzzing