Merge pull request #10417 from vespa-engine/toregge/use-unique-store-enumerator-when-saving-enumerated-attributes

Use unique store enumerator when saving enumerated attribute vectors.

4 files changed, 37 insertions, 5 deletions

diff --git a/vespalib/src/vespa/vespalib/datastore/datastorebase.cpp b/vespalib/src/vespa/vespalib/datastore/datastorebase.cpp
index 222e820546e..39c93dd98fe 100644
--- a/vespalib/src/vespa/vespalib/datastore/datastorebase.cpp
+++ b/vespalib/src/vespa/vespalib/datastore/datastorebase.cpp
@@ -84,6 +84,7 @@ DataStoreBase::DataStoreBase(uint32_t numBuffers, size_t maxArrays)
       _elemHold2List(),
       _numBuffers(numBuffers),
       _maxArrays(maxArrays),
+      _compaction_count(0u),
       _genHolder()
 {
 }
@@ -453,6 +454,7 @@ DataStoreBase::markCompacting(uint32_t bufferId)
     state.setCompacting();
     state.disableElemHoldList();
     state.setFreeListList(nullptr);
+    ++_compaction_count;
 }
 
 std::vector<uint32_t>
diff --git a/vespalib/src/vespa/vespalib/datastore/datastorebase.h b/vespalib/src/vespa/vespalib/datastore/datastorebase.h
index 4886237194f..2c7f3eae920 100644
--- a/vespalib/src/vespa/vespalib/datastore/datastorebase.h
+++ b/vespalib/src/vespa/vespalib/datastore/datastorebase.h
@@ -8,6 +8,7 @@
 #include <vespa/vespalib/util/memoryusage.h>
 #include <vector>
 #include <deque>
+#include <atomic>
 
 namespace search::datastore {
 
@@ -148,6 +149,7 @@ protected:
 
     const uint32_t _numBuffers;
     const size_t   _maxArrays;
+    mutable std::atomic<uint64_t> _compaction_count;
 
     vespalib::GenerationHolder _genHolder;
 
@@ -355,6 +357,8 @@ public:
 
     uint32_t startCompactWorstBuffer(uint32_t typeId);
     std::vector<uint32_t> startCompactWorstBuffers(bool compactMemory, bool compactAddressSpace);
+    uint64_t get_compaction_count() const { return _compaction_count.load(std::memory_order_relaxed); }
+    void bump_compaction_count() const { ++_compaction_count; }
 };
 
 }
diff --git a/vespalib/src/vespa/vespalib/datastore/unique_store_enumerator.h b/vespalib/src/vespa/vespalib/datastore/unique_store_enumerator.h
index 94c217bf836..a2e626c6b20 100644
--- a/vespalib/src/vespa/vespalib/datastore/unique_store_enumerator.h
+++ b/vespalib/src/vespa/vespalib/datastore/unique_store_enumerator.h
@@ -14,24 +14,29 @@ namespace search::datastore {
  */
 template <typename RefT>
 class UniqueStoreEnumerator {
+public:
     using RefType = RefT;
+    using EnumValues = std::vector<std::vector<uint32_t>>;
 
+private:
     const UniqueStoreDictionaryBase &_dict;
-    EntryRef _root;
+    EntryRef _frozen_root;
     const DataStoreBase &_store;
-    std::vector<std::vector<uint32_t>> _enumValues;
+    EnumValues _enumValues;
     uint32_t _next_enum_val;
 public:
     UniqueStoreEnumerator(const UniqueStoreDictionaryBase &dict, const DataStoreBase &store);
     ~UniqueStoreEnumerator();
+    EntryRef get_frozen_root() const { return _frozen_root; }
     void enumerateValue(EntryRef ref);
     void enumerateValues();
+    void clear();
 
     template <typename Function>
     void
     foreach_key(Function &&func) const
     {
-        _dict.foreach_key(_root, func);
+        _dict.foreach_key(_frozen_root, func);
     }
 
     uint32_t mapEntryRefToEnumValue(EntryRef ref) const {
@@ -45,6 +50,19 @@ public:
             return 0u;
         }
     }
+
+    uint32_t map_entry_ref_to_enum_value_or_zero(EntryRef ref) const {
+        if (ref.valid()) {
+            RefType iRef(ref);
+            if (iRef.unscaled_offset() < _enumValues[iRef.bufferId()].size()) {
+                return  _enumValues[iRef.bufferId()][iRef.unscaled_offset()];
+            } else {
+                return 0u;
+            }
+        } else {
+            return 0u;
+        }
+    }
 };
 
 }
diff --git a/vespalib/src/vespa/vespalib/datastore/unique_store_enumerator.hpp b/vespalib/src/vespa/vespalib/datastore/unique_store_enumerator.hpp
index b364c2a2ba3..a6053d4f64e 100644
--- a/vespalib/src/vespa/vespalib/datastore/unique_store_enumerator.hpp
+++ b/vespalib/src/vespa/vespalib/datastore/unique_store_enumerator.hpp
@@ -9,8 +9,9 @@ namespace search::datastore {
 template <typename RefT>
 UniqueStoreEnumerator<RefT>::UniqueStoreEnumerator(const UniqueStoreDictionaryBase &dict, const DataStoreBase &store)
     : _dict(dict),
-      _root(_dict.get_frozen_root()),
+      _frozen_root(_dict.get_frozen_root()),
       _store(store),
+      _enumValues(),
       _next_enum_val(1)
 {
 }
@@ -45,7 +46,14 @@ UniqueStoreEnumerator<RefT>::enumerateValues()
         }
     }
     _next_enum_val = 1;
-    _dict.foreach_key(_root, [this](EntryRef ref) { enumerateValue(ref); });
+    _dict.foreach_key(_frozen_root, [this](EntryRef ref) { enumerateValue(ref); });
+}
+
+template <typename RefT>
+void
+UniqueStoreEnumerator<RefT>::clear()
+{
+    EnumValues().swap(_enumValues);
 }
 
 }