From 9e97d5cf9abefbe14edd734be07eb3a51ac0bca9 Mon Sep 17 00:00:00 2001
From: Marcus Holland-Moritz <github@mhxnet.de>
Date: Mon, 18 Aug 2025 14:50:33 +0200
Subject: [PATCH] feat(mkdwarfs): add `--no-metadata-version-history`

---
 doc/mkdwarfs.md                          |  8 +++++
 include/dwarfs/writer/metadata_options.h |  1 +
 src/writer/internal/metadata_builder.cpp |  8 +++--
 test/tool_main_test.cpp                  | 37 ++++++++++++++++++++++--
 tools/src/mkdwarfs_main.cpp              |  3 ++
 5 files changed, 53 insertions(+), 4 deletions(-)

diff --git a/doc/mkdwarfs.md b/doc/mkdwarfs.md
index facd6230..9daba13f 100644
--- a/doc/mkdwarfs.md
+++ b/doc/mkdwarfs.md
@@ -252,6 +252,14 @@ Most other options are concerned with compression tuning:
   is unchanged, this will still re-order and re-compress *all* blocks. Implies
   `--recompress=all` and `--rebuild-metadata`.
 
+- `--no-metadata-version-history`:
+  By default, when rebuilding the metadata block, a small history entry will
+  be added to the metadata block to keep track of versions and options across
+  metadata rebuilds. It is recommended to keep this history, as it allows for
+  better understanding of how the metadata block has changed over time and
+  why it may be lacking certain features. If you know what you're doing, you
+  can use this option to disable this history completely.
+
 - `--recompress-categories=`[`!`]*category*[`,`...]:
   When `--recompress` is set to `all` or `block`, this option controls
   which categories of blocks will be recompressed. Adding a `!` in front
diff --git a/include/dwarfs/writer/metadata_options.h b/include/dwarfs/writer/metadata_options.h
index 29c657b2..37a78cf8 100644
--- a/include/dwarfs/writer/metadata_options.h
+++ b/include/dwarfs/writer/metadata_options.h
@@ -56,6 +56,7 @@ struct metadata_options {
   bool no_create_timestamp{false};
   bool no_category_names{false};
   bool no_category_metadata{false};
+  bool no_metadata_version_history{false};
   size_t inode_size_cache_min_chunk_count{128};
 };
 
diff --git a/src/writer/internal/metadata_builder.cpp b/src/writer/internal/metadata_builder.cpp
index e4dc3bc4..7b060cac 100644
--- a/src/writer/internal/metadata_builder.cpp
+++ b/src/writer/internal/metadata_builder.cpp
@@ -772,8 +772,12 @@ void metadata_builder_<LoggerPolicy>::upgrade_metadata(
 
   tv << "upgrading metadata...";
 
-  md_.metadata_version_history().ensure();
-  md_.metadata_version_history()->push_back(std::move(histent));
+  if (options_.no_metadata_version_history) {
+    md_.metadata_version_history().reset();
+  } else {
+    md_.metadata_version_history().ensure();
+    md_.metadata_version_history()->push_back(std::move(histent));
+  }
 }
 
 } // namespace
diff --git a/test/tool_main_test.cpp b/test/tool_main_test.cpp
index 398eedd0..031dc977 100644
--- a/test/tool_main_test.cpp
+++ b/test/tool_main_test.cpp
@@ -4104,9 +4104,15 @@ TEST(mkdwarfs_test, change_block_size) {
 
       auto fs = t.fs_from_stdout();
       auto info =
-          fs.info_as_json({.features = reader::fsinfo_features::for_level(2)});
+          fs.info_as_json({.features = reader::fsinfo_features::for_level(3)});
 
       EXPECT_EQ(1 << lg_block_size, info["block_size"].get<int>());
+
+      auto const& hist = info["meta"]["metadata_version_history"];
+
+      ASSERT_EQ(1, hist.size());
+
+      EXPECT_EQ(1 << 18, hist[0]["block_size"].get<int>());
     }
 
     auto t2 = rebuild_tester(t.out());
@@ -4121,9 +4127,36 @@ TEST(mkdwarfs_test, change_block_size) {
 
       auto fs = t2.fs_from_stdout();
       auto info =
-          fs.info_as_json({.features = reader::fsinfo_features::for_level(2)});
+          fs.info_as_json({.features = reader::fsinfo_features::for_level(3)});
 
       EXPECT_EQ(1 << 18, info["block_size"].get<int>());
+
+      auto const& hist = info["meta"]["metadata_version_history"];
+
+      ASSERT_EQ(2, hist.size());
+
+      EXPECT_EQ(1 << 18, hist[0]["block_size"].get<int>());
+      EXPECT_EQ(1 << lg_block_size, hist[1]["block_size"].get<int>());
+    }
+
+    if (lg_block_size == 10) {
+      auto t3 = rebuild_tester(t2.out());
+      ASSERT_EQ(
+          0, t3.run({"-i", image_file, "-o", "-", "-S20", "-C", "zstd:level=5",
+                     "--change-block-size", "--keep-all-times",
+                     "--log-level=debug", "--no-metadata-version-history"}))
+          << t3.err();
+
+      {
+        auto fs = t3.fs_from_stdout();
+        auto info = fs.info_as_json(
+            {.features = reader::fsinfo_features::for_level(3)});
+
+        EXPECT_EQ(1 << 20, info["block_size"].get<int>());
+
+        EXPECT_FALSE(info["meta"].contains("metadata_version_history"))
+            << info.dump(2);
+      }
     }
   }
 }
diff --git a/tools/src/mkdwarfs_main.cpp b/tools/src/mkdwarfs_main.cpp
index ff41d418..c671cd0a 100644
--- a/tools/src/mkdwarfs_main.cpp
+++ b/tools/src/mkdwarfs_main.cpp
@@ -534,6 +534,9 @@ int mkdwarfs_main(int argc, sys_char** argv, iolayer const& iol) {
     ("change-block-size",
         po::value<bool>(&change_block_size)->zero_tokens(),
         "change block size when recompressing")
+    ("no-metadata-version-history",
+        po::value<bool>(&options.metadata.no_metadata_version_history)->zero_tokens(),
+        "remove metadata version history")
     ("recompress-categories",
         po::value<std::string>(&recompress_categories),
         "only recompress blocks of these categories")