Optimize handling of CLEARDOC updates by caching the entry ref for the default value once.

6 files changed, 23 insertions, 0 deletions

diff --git a/searchlib/src/vespa/searchlib/attribute/changevector.h b/searchlib/src/vespa/searchlib/attribute/changevector.h
index 014c218bc76..271497398a8 100644
--- a/searchlib/src/vespa/searchlib/attribute/changevector.h
+++ b/searchlib/src/vespa/searchlib/attribute/changevector.h
@@ -46,6 +46,7 @@ struct ChangeBase {
     uint32_t get_entry_ref() const { return _cached_entry_ref; }
     void set_entry_ref(uint32_t entry_ref) const { _cached_entry_ref = entry_ref; }
     bool has_entry_ref() const { return _cached_entry_ref != UNSET_ENTRY_REF; }
+    void clear_entry_ref() const { _cached_entry_ref = UNSET_ENTRY_REF; }
 
     Type               _type;
     uint32_t           _doc;
diff --git a/searchlib/src/vespa/searchlib/attribute/enumattribute.h b/searchlib/src/vespa/searchlib/attribute/enumattribute.h
index 8136e654152..f0ff23a06b4 100644
--- a/searchlib/src/vespa/searchlib/attribute/enumattribute.h
+++ b/searchlib/src/vespa/searchlib/attribute/enumattribute.h
@@ -56,6 +56,7 @@ protected:
     virtual void considerAttributeChange(const Change & c, EnumStoreBatchUpdater & inserter) = 0;
     vespalib::MemoryUsage getEnumStoreValuesMemoryUsage() const override;
     void populate_address_space_usage(AddressSpaceUsage& usage) const override;
+    void cache_change_data_entry_ref(const Change& c) const;
 public:
     EnumAttribute(const vespalib::string & baseFileName, const AttributeVector::Config & cfg);
     ~EnumAttribute();
diff --git a/searchlib/src/vespa/searchlib/attribute/enumattribute.hpp b/searchlib/src/vespa/searchlib/attribute/enumattribute.hpp
index 3e578856c2b..c0680fd9238 100644
--- a/searchlib/src/vespa/searchlib/attribute/enumattribute.hpp
+++ b/searchlib/src/vespa/searchlib/attribute/enumattribute.hpp
@@ -84,6 +84,15 @@ EnumAttribute<B>::populate_address_space_usage(AddressSpaceUsage& usage) const
     usage.set(AddressSpaceComponents::enum_store, _enumStore.get_values_address_space_usage());
 }
 
+template <typename B>
+void
+EnumAttribute<B>::cache_change_data_entry_ref(const Change& c) const
+{
+    EnumIndex new_idx;
+    _enumStore.find_index(c._data.raw(), new_idx);
+    c.set_entry_ref(new_idx.ref());
+}
+
 } // namespace search
 
 
diff --git a/searchlib/src/vespa/searchlib/attribute/singleenumattribute.hpp b/searchlib/src/vespa/searchlib/attribute/singleenumattribute.hpp
index dc6db1d5a82..a51c9804cf2 100644
--- a/searchlib/src/vespa/searchlib/attribute/singleenumattribute.hpp
+++ b/searchlib/src/vespa/searchlib/attribute/singleenumattribute.hpp
@@ -181,6 +181,8 @@ void
 SingleValueEnumAttribute<B>::applyValueChanges(EnumStoreBatchUpdater& updater)
 {
     ValueModifier valueGuard(this->getValueModifier());
+    // This avoids searching for the defaultValue in the enum store for each CLEARDOC in the change vector.
+    this->cache_change_data_entry_ref(this->_defaultValue);
     for (const auto& change : this->_changes.getInsertOrder()) {
         if (change._type == ChangeBase::UPDATE) {
             applyUpdateValueChange(change, updater);
@@ -192,6 +194,8 @@ SingleValueEnumAttribute<B>::applyValueChanges(EnumStoreBatchUpdater& updater)
             applyUpdateValueChange(clearDoc, updater);
         }
     }
+    // We must clear the cached entry ref as the defaultValue might be located in another data buffer on later invocations.
+    this->_defaultValue.clear_entry_ref();
 }
 
 template <typename B>
diff --git a/searchlib/src/vespa/searchlib/attribute/singlenumericpostattribute.hpp b/searchlib/src/vespa/searchlib/attribute/singlenumericpostattribute.hpp
index 4d4e4619d51..4d37171e151 100644
--- a/searchlib/src/vespa/searchlib/attribute/singlenumericpostattribute.hpp
+++ b/searchlib/src/vespa/searchlib/attribute/singlenumericpostattribute.hpp
@@ -89,6 +89,8 @@ SingleValueNumericPostingAttribute<B>::applyValueChanges(EnumStoreBatchUpdater&
     // used to make sure several arithmetic operations on the same document in a single commit works
     std::map<DocId, EnumIndex> currEnumIndices;
 
+    // This avoids searching for the defaultValue in the enum store for each CLEARDOC in the change vector.
+    this->cache_change_data_entry_ref(this->_defaultValue);
     for (const auto& change : this->_changes.getInsertOrder()) {
         auto enumIter = currEnumIndices.find(change._doc);
         EnumIndex oldIdx;
@@ -114,6 +116,8 @@ SingleValueNumericPostingAttribute<B>::applyValueChanges(EnumStoreBatchUpdater&
             applyUpdateValueChange(clearDoc, enumStore, currEnumIndices);
         }
     }
+    // We must clear the cached entry ref as the defaultValue might be located in another data buffer on later invocations.
+    this->_defaultValue.clear_entry_ref();
 
     makePostingChange(enumStore.get_comparator(), currEnumIndices, changePost);
 
diff --git a/searchlib/src/vespa/searchlib/attribute/singlestringpostattribute.hpp b/searchlib/src/vespa/searchlib/attribute/singlestringpostattribute.hpp
index 592ecb68b1e..4670ee075fe 100644
--- a/searchlib/src/vespa/searchlib/attribute/singlestringpostattribute.hpp
+++ b/searchlib/src/vespa/searchlib/attribute/singlestringpostattribute.hpp
@@ -90,6 +90,8 @@ SingleValueStringPostingAttributeT<B>::applyValueChanges(EnumStoreBatchUpdater&
     // used to make sure several arithmetic operations on the same document in a single commit works
     std::map<DocId, EnumIndex> currEnumIndices;
 
+    // This avoids searching for the defaultValue in the enum store for each CLEARDOC in the change vector.
+    this->cache_change_data_entry_ref(this->_defaultValue);
     for (const auto& change : this->_changes.getInsertOrder()) {
         auto enumIter = currEnumIndices.find(change._doc);
         EnumIndex oldIdx;
@@ -105,6 +107,8 @@ SingleValueStringPostingAttributeT<B>::applyValueChanges(EnumStoreBatchUpdater&
             applyUpdateValueChange(this->_defaultValue, enumStore, currEnumIndices);
         }
     }
+    // We must clear the cached entry ref as the defaultValue might be located in another data buffer on later invocations.
+    this->_defaultValue.clear_entry_ref();
 
     makePostingChange(enumStore.get_folded_comparator(), dictionary, currEnumIndices, changePost);