From 201b6ddb3bcadb6f1c965acf083e1bdaa9bda04b Mon Sep 17 00:00:00 2001
From: Marcus Holland-Moritz <github@mhxnet.de>
Date: Wed, 30 Dec 2020 13:24:50 +0100
Subject: [PATCH] Dynamically adapt nilsimsa search depth

---
 doc/mkdwarfs.md                    | 14 ++++--
 include/dwarfs/console_writer.h    |  1 +
 include/dwarfs/filesystem_writer.h |  3 ++
 include/dwarfs/inode_manager.h     |  8 +--
 include/dwarfs/options.h           |  1 +
 include/dwarfs/progress.h          |  3 ++
 src/dwarfs/console_writer.cpp      | 26 +++++++++-
 src/dwarfs/filesystem_writer.cpp   |  1 +
 src/dwarfs/inode_manager.cpp       | 63 ++++++++++++++++--------
 src/dwarfs/scanner.cpp             | 10 +++-
 src/mkdwarfs.cpp                   | 78 ++++++++++++++++++++----------
 11 files changed, 152 insertions(+), 56 deletions(-)

diff --git a/doc/mkdwarfs.md b/doc/mkdwarfs.md
index aafd65bf..42ffa2d2 100644
--- a/doc/mkdwarfs.md
+++ b/doc/mkdwarfs.md
@@ -154,7 +154,7 @@ Most other options are concerned with compression tuning:
     the `mtime` field in order to save metadata space. If you want to save
     `atime` and `ctime` as well, use this option.
 
-  * `--order=none`|`path`|`similarity`|`nilsimsa`[`:`*limit*[`:`*depth*]]|`script`:
+  * `--order=none`|`path`|`similarity`|`nilsimsa`[`:`*limit*[`:`*depth*[`:`*mindepth*]]]|`script`:
     The order in which inodes will be written to the file system. Choosing `none`,
     the inodes will be stored in the order in which they are discovered. With
     `path`, they will be sorted asciibetically by path name of the first file
@@ -168,9 +168,15 @@ Most other options are concerned with compression tuning:
     enough for adding. A *limit* of 255 means "essentially identical", whereas
     a *limit* of 0 means "not similar at all". The *depth* determines up to
     how many inodes can be checked at most while searching for a similar one.
-    The default if you omit these values is a *limit* of 255 and a *depth*
-    of 20000. Last but not least, if scripting support is built into `mkdwarfs`,
-    you can choose `script` to let the script determine the order.
+    To avoid nilsimsa ordering to become a bottleneck when ordering lots of
+    small files, the *depth* is adjusted dynamically to keep the input queue
+    to the segmentation/compression stages adequately filled. You can specify
+    how much the *depth* can be adjusted by also specifying *mindepth*.
+    The default if you omit these values is a *limit* of 255, a *depth*
+    of 20000 and a *mindepth* of 1000. Note that if you want reproducible
+    results, you need to set *depth* and *mindepth* to the same value.
+    Last but not least, if scripting support is built into `mkdwarfs`, you can
+    choose `script` to let the script determine the order.
 
   * `--blockhash-window-sizes=`*value*[,*value*]...:
     Window sizes used for block hashing. These sizes, separated by commas,
diff --git a/include/dwarfs/console_writer.h b/include/dwarfs/console_writer.h
index 04d91147..48f9c8c5 100644
--- a/include/dwarfs/console_writer.h
+++ b/include/dwarfs/console_writer.h
@@ -61,5 +61,6 @@ class console_writer : public logger {
   display_mode const mode_;
   bool const color_;
   bool const with_context_;
+  bool const debug_progress_;
 };
 } // namespace dwarfs
diff --git a/include/dwarfs/filesystem_writer.h b/include/dwarfs/filesystem_writer.h
index 5ce4883e..4445d910 100644
--- a/include/dwarfs/filesystem_writer.h
+++ b/include/dwarfs/filesystem_writer.h
@@ -87,6 +87,8 @@ class filesystem_writer {
 
   size_t size() const { return impl_->size(); }
 
+  int queue_fill() const { return impl_->queue_fill(); }
+
   class impl {
    public:
     virtual ~impl() = default;
@@ -99,6 +101,7 @@ class filesystem_writer {
                              folly::ByteRange data) = 0;
     virtual void flush() = 0;
     virtual size_t size() const = 0;
+    virtual int queue_fill() const = 0;
   };
 
  private:
diff --git a/include/dwarfs/inode_manager.h b/include/dwarfs/inode_manager.h
index a807681a..f840fccb 100644
--- a/include/dwarfs/inode_manager.h
+++ b/include/dwarfs/inode_manager.h
@@ -29,6 +29,7 @@ namespace dwarfs {
 
 class inode;
 class logger;
+class progress;
 class script;
 
 struct file_order_options;
@@ -36,8 +37,9 @@ struct file_order_options;
 class inode_manager {
  public:
   using inode_cb = std::function<void(std::shared_ptr<inode> const&)>;
+  using order_cb = std::function<int64_t(std::shared_ptr<inode> const&)>;
 
-  inode_manager(logger& lgr);
+  inode_manager(logger& lgr, progress& prog);
 
   std::shared_ptr<inode> create_inode() { return impl_->create_inode(); }
 
@@ -45,7 +47,7 @@ class inode_manager {
 
   void order_inodes(std::shared_ptr<script> scr,
                     file_order_options const& file_order, uint32_t first_inode,
-                    inode_cb const& fn) {
+                    order_cb const& fn) {
     impl_->order_inodes(std::move(scr), file_order, first_inode, fn);
   }
 
@@ -59,7 +61,7 @@ class inode_manager {
     virtual size_t count() const = 0;
     virtual void order_inodes(std::shared_ptr<script> scr,
                               file_order_options const& file_order,
-                              uint32_t first_inode, inode_cb const& fn) = 0;
+                              uint32_t first_inode, order_cb const& fn) = 0;
     virtual void for_each_inode(
         std::function<void(std::shared_ptr<inode> const&)> const& fn) const = 0;
   };
diff --git a/include/dwarfs/options.h b/include/dwarfs/options.h
index 26ed2ca6..c8c6237e 100644
--- a/include/dwarfs/options.h
+++ b/include/dwarfs/options.h
@@ -59,6 +59,7 @@ enum class file_order_mode { NONE, PATH, SCRIPT, SIMILARITY, NILSIMSA };
 struct file_order_options {
   file_order_mode mode{file_order_mode::NONE};
   int nilsimsa_depth{20000};
+  int nilsimsa_min_depth{1000};
   int nilsimsa_limit{255};
 };
 
diff --git a/include/dwarfs/progress.h b/include/dwarfs/progress.h
index 27e4525b..d094b1eb 100644
--- a/include/dwarfs/progress.h
+++ b/include/dwarfs/progress.h
@@ -70,6 +70,9 @@ class progress {
   std::atomic<size_t> inodes_written{0};
   std::atomic<size_t> blocks_written{0};
   std::atomic<size_t> errors{0};
+  std::atomic<size_t> nilsimsa_depth{0};
+  std::atomic<size_t> blockify_queue{0};
+  std::atomic<size_t> compress_queue{0};
   std::atomic<uint64_t> original_size{0};
   std::atomic<uint64_t> hardlink_size{0};
   std::atomic<uint64_t> saved_by_deduplication{0};
diff --git a/src/dwarfs/console_writer.cpp b/src/dwarfs/console_writer.cpp
index 27f810b6..bd454fa7 100644
--- a/src/dwarfs/console_writer.cpp
+++ b/src/dwarfs/console_writer.cpp
@@ -19,6 +19,7 @@
  * along with dwarfs.  If not, see <https://www.gnu.org/licenses/>.
  */
 
+#include <cstdlib>
 #include <cstring>
 #include <locale>
 #include <sstream>
@@ -44,6 +45,15 @@ namespace {
 char const* const asc_bar[8] = {"=", "=", "=", "=", "=", "=", "=", "="};
 char const* const uni_bar[8] = {"▏", "▎", "▍", "▌", "▋", "▊", "▉", "█"};
 
+bool is_debug_progress() {
+  if (auto var = ::getenv("DWARFS_DEBUG_PROGRESS")) {
+    if (auto val = folly::tryTo<bool>(var)) {
+      return *val;
+    }
+  }
+  return false;
+}
+
 } // namespace
 
 console_writer::console_writer(std::ostream& os, progress_mode pg_mode,
@@ -56,7 +66,8 @@ console_writer::console_writer(std::ostream& os, progress_mode pg_mode,
     , width_(width)
     , mode_(mode)
     , color_(stream_is_fancy_terminal(os))
-    , with_context_(with_context) {
+    , with_context_(with_context)
+    , debug_progress_(is_debug_progress()) {
   os_.imbue(std::locale(os_.getloc(),
                         new boost::posix_time::time_facet("%H:%M:%S.%f")));
   if (threshold > level_type::INFO) {
@@ -176,7 +187,18 @@ void console_writer::update(const progress& p, bool last) {
           << p.files_found - p.duplicate_files << " inodes)" << newline
 
           << "compressed filesystem: " << p.blocks_written << " blocks/"
-          << size_with_unit(p.compressed_size) << " written" << newline;
+          << size_with_unit(p.compressed_size) << " written";
+
+      if (debug_progress_) {
+        oss << " [" << p.nilsimsa_depth << "/" << p.blockify_queue << "/"
+            << p.compress_queue << "]";
+      } else {
+        if (p.nilsimsa_depth > 0) {
+          oss << " [depth: " << p.nilsimsa_depth << "]";
+        }
+      }
+
+      oss << newline;
       break;
 
     case REWRITE:
diff --git a/src/dwarfs/filesystem_writer.cpp b/src/dwarfs/filesystem_writer.cpp
index 76dba7b3..e1f4ccc3 100644
--- a/src/dwarfs/filesystem_writer.cpp
+++ b/src/dwarfs/filesystem_writer.cpp
@@ -218,6 +218,7 @@ class filesystem_writer_ : public filesystem_writer::impl {
                                 folly::ByteRange data) override;
   void flush() override;
   size_t size() const override { return os_.tellp(); }
+  int queue_fill() const override { return static_cast<int>(wg_.queue_size()); }
 
  private:
   void write_section(section_type type, std::vector<uint8_t>&& data,
diff --git a/src/dwarfs/inode_manager.cpp b/src/dwarfs/inode_manager.cpp
index cccf1a29..9412c74f 100644
--- a/src/dwarfs/inode_manager.cpp
+++ b/src/dwarfs/inode_manager.cpp
@@ -37,6 +37,7 @@
 #include "dwarfs/nilsimsa.h"
 #include "dwarfs/options.h"
 #include "dwarfs/os_access.h"
+#include "dwarfs/progress.h"
 #include "dwarfs/script.h"
 #include "dwarfs/similarity.h"
 
@@ -149,8 +150,9 @@ class nilsimsa_cache_entry {
 template <typename LoggerPolicy>
 class inode_manager_ : public inode_manager::impl {
  public:
-  inode_manager_(logger& lgr)
-      : log_(lgr) {}
+  inode_manager_(logger& lgr, progress& prog)
+      : LOG_PROXY_INIT(lgr)
+      , prog_(prog) {}
 
   std::shared_ptr<inode> create_inode() override {
     auto ino = std::make_shared<inode_>();
@@ -162,7 +164,7 @@ class inode_manager_ : public inode_manager::impl {
 
   void order_inodes(std::shared_ptr<script> scr,
                     file_order_options const& file_order, uint32_t first_inode,
-                    inode_manager::inode_cb const& fn) override;
+                    inode_manager::order_cb const& fn) override;
 
   void
   for_each_inode(std::function<void(std::shared_ptr<inode> const&)> const& fn)
@@ -215,7 +217,7 @@ class inode_manager_ : public inode_manager::impl {
                      std::vector<uint32_t>& index);
 
   void
-  order_inodes_by_nilsimsa(inode_manager::inode_cb const& fn, uint32_t inode_no,
+  order_inodes_by_nilsimsa(inode_manager::order_cb const& fn, uint32_t inode_no,
                            file_order_options const& file_order);
 
   void number_inodes(size_t first_no) {
@@ -225,13 +227,14 @@ class inode_manager_ : public inode_manager::impl {
   }
 
   std::vector<std::shared_ptr<inode>> inodes_;
-  log_proxy<LoggerPolicy> log_;
+  LOG_PROXY_DECL(LoggerPolicy);
+  progress& prog_;
 };
 
 template <typename LoggerPolicy>
 void inode_manager_<LoggerPolicy>::order_inodes(
     std::shared_ptr<script> scr, file_order_options const& file_order,
-    uint32_t first_inode, inode_manager::inode_cb const& fn) {
+    uint32_t first_inode, inode_manager::order_cb const& fn) {
   switch (file_order.mode) {
   case file_order_mode::NONE:
     LOG_INFO << "keeping inode order";
@@ -276,7 +279,9 @@ void inode_manager_<LoggerPolicy>::order_inodes(
 
   LOG_INFO << "assigning file inodes...";
   number_inodes(first_inode);
-  for_each_inode(fn);
+  for (const auto& ino : inodes_) {
+    fn(ino);
+  }
 }
 
 template <typename LoggerPolicy>
@@ -322,7 +327,7 @@ void inode_manager_<LoggerPolicy>::presort_index(
 
 template <typename LoggerPolicy>
 void inode_manager_<LoggerPolicy>::order_inodes_by_nilsimsa(
-    inode_manager::inode_cb const& fn, uint32_t inode_no,
+    inode_manager::order_cb const& fn, uint32_t inode_no,
     file_order_options const& file_order) {
   auto count = inodes_.size();
 
@@ -340,7 +345,7 @@ void inode_manager_<LoggerPolicy>::order_inodes_by_nilsimsa(
     inodes_.push_back(std::move(inodes[index.back()]));
     index.pop_back();
     inodes_.back()->set_num(inode_no++);
-    fn(inodes_.back());
+    return fn(inodes_.back());
   };
 
   if (empty != index.end()) {
@@ -349,10 +354,15 @@ void inode_manager_<LoggerPolicy>::order_inodes_by_nilsimsa(
   }
 
   if (!index.empty()) {
-    const int depth = file_order.nilsimsa_depth;
-    const int limit = file_order.nilsimsa_limit;
+    const int_fast32_t max_depth = file_order.nilsimsa_depth;
+    const int_fast32_t min_depth =
+        std::min<int32_t>(file_order.nilsimsa_min_depth, max_depth);
+    const int_fast32_t limit = file_order.nilsimsa_limit;
+    int_fast32_t depth = max_depth;
+    int64_t processed = 0;
 
-    LOG_INFO << "nilsimsa: depth=" << depth << ", limit=" << limit;
+    LOG_INFO << "nilsimsa: depth=" << depth << " (" << min_depth
+             << "), limit=" << limit;
 
     presort_index(inodes, index);
 
@@ -361,12 +371,12 @@ void inode_manager_<LoggerPolicy>::order_inodes_by_nilsimsa(
     while (!index.empty()) {
       auto* ref_hash = inodes_.back()->nilsimsa_similarity_hash().data();
 
-      int max_sim = 0;
-      int max_sim_ix = 0;
+      int_fast32_t max_sim = 0;
+      int_fast32_t max_sim_ix = 0;
 
-      int end = int(index.size()) > depth ? index.size() - depth : 0;
+      int_fast32_t end = int(index.size()) > depth ? index.size() - depth : 0;
 
-      for (int i = index.size() - 1; i >= end; --i) {
+      for (int_fast32_t i = index.size() - 1; i >= end; --i) {
         auto sim = dwarfs::nilsimsa_similarity(
             ref_hash, inodes[index[i]]->nilsimsa_similarity_hash().data());
 
@@ -385,7 +395,22 @@ void inode_manager_<LoggerPolicy>::order_inodes_by_nilsimsa(
       std::rotate(index.begin() + max_sim_ix, index.begin() + max_sim_ix + 1,
                   index.end());
 
-      finalize_inode();
+      auto fill = finalize_inode();
+
+      if (++processed >= 4096 && processed % 32 == 0) {
+        constexpr int64_t smooth = 512;
+        auto target_depth = fill * max_depth / 2048;
+
+        depth = ((smooth - 1) * depth + target_depth) / smooth;
+
+        if (depth > max_depth) {
+          depth = max_depth;
+        } else if (depth < min_depth) {
+          depth = min_depth;
+        }
+      }
+
+      prog_.nilsimsa_depth = depth;
     }
   }
 
@@ -394,8 +419,8 @@ void inode_manager_<LoggerPolicy>::order_inodes_by_nilsimsa(
   }
 }
 
-inode_manager::inode_manager(logger& lgr)
+inode_manager::inode_manager(logger& lgr, progress& prog)
     : impl_(make_unique_logging_object<impl, inode_manager_, logger_policies>(
-          lgr)) {}
+          lgr, prog)) {}
 
 } // namespace dwarfs
diff --git a/src/dwarfs/scanner.cpp b/src/dwarfs/scanner.cpp
index 7c7cb33a..eb25334a 100644
--- a/src/dwarfs/scanner.cpp
+++ b/src/dwarfs/scanner.cpp
@@ -485,7 +485,7 @@ void scanner_<LoggerPolicy>::scan(filesystem_writer& fsw,
 
   LOG_INFO << "finding duplicate files...";
 
-  inode_manager im(lgr_);
+  inode_manager im(lgr_, prog);
 
   file_deduplication_visitor fdv;
   root->accept(fdv);
@@ -540,7 +540,7 @@ void scanner_<LoggerPolicy>::scan(filesystem_writer& fsw,
   LOG_INFO << "building blocks...";
   block_manager bm(lgr_, prog, cfg_, os_, fsw);
 
-  worker_group blockify("blockify");
+  worker_group blockify("blockify", 1, 1 << 20);
 
   im.order_inodes(script_, options_.file_order, first_file_inode,
                   [&](std::shared_ptr<inode> const& ino) {
@@ -549,6 +549,12 @@ void scanner_<LoggerPolicy>::scan(filesystem_writer& fsw,
                       bm.add_inode(ino);
                       prog.inodes_written++;
                     });
+                    auto queued_files = blockify.queue_size();
+                    auto queued_blocks = fsw.queue_fill();
+                    prog.blockify_queue = queued_files;
+                    prog.compress_queue = queued_blocks;
+                    return INT64_C(500) * queued_blocks +
+                           static_cast<int64_t>(queued_files);
                   });
 
   LOG_INFO << "waiting for segmenting/blockifying to finish...";
diff --git a/src/mkdwarfs.cpp b/src/mkdwarfs.cpp
index 3b31e198..b581a1de 100644
--- a/src/mkdwarfs.cpp
+++ b/src/mkdwarfs.cpp
@@ -30,6 +30,7 @@
 #include <iterator>
 #include <map>
 #include <memory>
+#include <optional>
 #include <sstream>
 #include <stdexcept>
 #include <string>
@@ -171,6 +172,40 @@ class script_options : public options_interface {
 
 namespace {
 
+int parse_order_option(std::string const& ordname, std::string const& opt,
+                       int& value, std::string_view name,
+                       std::optional<int> min = std::nullopt,
+                       std::optional<int> max = std::nullopt) {
+  if (!opt.empty()) {
+    if (auto val = folly::tryTo<int>(opt)) {
+      auto tmp = *val;
+      if (min && max && (tmp < *min || tmp > *max)) {
+        std::cerr << "error: " << name << " (" << opt
+                  << ") out of range for order '" << ordname << "' (" << *min
+                  << ".." << *max << ")" << std::endl;
+        return 1;
+      }
+      if (min && tmp < *min) {
+        std::cerr << "error: " << name << " (" << opt
+                  << ") cannot be less than " << *min << " for order '"
+                  << ordname << "'" << std::endl;
+      }
+      if (max && tmp > *max) {
+        std::cerr << "error: " << name << " (" << opt
+                  << ") cannot be greater than " << *max << " for order '"
+                  << ordname << "'" << std::endl;
+      }
+      value = tmp;
+    } else {
+      std::cerr << "error: " << name << " (" << opt
+                << ") is not numeric for order '" << ordname << "'"
+                << std::endl;
+      return 1;
+    }
+  }
+  return 0;
+}
+
 size_t get_term_width() {
   struct ::winsize w;
   ::ioctl(STDOUT_FILENO, TIOCGWINSZ, &w);
@@ -281,7 +316,7 @@ constexpr std::array<level_defaults, 10> levels{{
     /* 6 */ {24, ALG_DATA_6, ALG_SCHEMA, "null",         "16,14,12", "nilsimsa"},
     /* 7 */ {24, ALG_DATA_7, ALG_SCHEMA, ALG_METADATA_7, "16,14,12", "nilsimsa"},
     /* 8 */ {24, ALG_DATA_8, ALG_SCHEMA, ALG_METADATA_9, "16,14,12", "nilsimsa"},
-    /* 9 */ {26, ALG_DATA_9, ALG_SCHEMA, ALG_METADATA_9, "16,14,12", "nilsimsa:255:50000"},
+    /* 9 */ {26, ALG_DATA_9, ALG_SCHEMA, ALG_METADATA_9, "16,14,12", "nilsimsa::50000"},
     // clang-format on
 }};
 
@@ -548,40 +583,31 @@ int mkdwarfs(int argc, char** argv) {
         return 1;
       }
 
-      if (order_opts.size() > 3) {
+      if (order_opts.size() > 4) {
         std::cerr << "error: too many options for inode order mode '"
                   << order_opts[0] << "'" << std::endl;
         return 1;
       }
 
-      if (auto val = folly::tryTo<int>(order_opts[1])) {
-        options.file_order.nilsimsa_limit = *val;
-        if (options.file_order.nilsimsa_limit < 0 ||
-            options.file_order.nilsimsa_limit > 255) {
-          std::cerr << "error: limit (" << order_opts[1]
-                    << ") out of range for order '" << order_opts[0]
-                    << "' (0..255)" << std::endl;
-          return 1;
-        }
-      } else {
-        std::cerr << "error: limit (" << order_opts[1]
-                  << ") is not numeric for order '" << order_opts[0] << "'"
-                  << std::endl;
+      auto ordname = order_opts[0];
+
+      if (parse_order_option(ordname, order_opts[1],
+                             options.file_order.nilsimsa_limit, "limit", 0,
+                             255)) {
         return 1;
       }
 
       if (order_opts.size() > 2) {
-        if (auto val = folly::tryTo<int>(order_opts[2])) {
-          options.file_order.nilsimsa_depth = *val;
-          if (options.file_order.nilsimsa_depth < 0) {
-            std::cerr << "error: depth (" << order_opts[2]
-                      << ") cannot be negative for order '" << order_opts[0]
-                      << "'" << std::endl;
-          }
-        } else {
-          std::cerr << "error: depth (" << order_opts[2]
-                    << ") is not numeric for order '" << order_opts[0] << "'"
-                    << std::endl;
+        if (parse_order_option(ordname, order_opts[2],
+                               options.file_order.nilsimsa_depth, "depth", 0)) {
+          return 1;
+        }
+      }
+
+      if (order_opts.size() > 3) {
+        if (parse_order_option(ordname, order_opts[3],
+                               options.file_order.nilsimsa_min_depth,
+                               "min depth", 0)) {
           return 1;
         }
       }