Propagate memory allocator to attribute vector components.

9 files changed, 100 insertions, 7 deletions

diff --git a/searchlib/src/tests/attribute/attribute_test.cpp b/searchlib/src/tests/attribute/attribute_test.cpp
index a26e751ae73..0ae495035f0 100644
--- a/searchlib/src/tests/attribute/attribute_test.cpp
+++ b/searchlib/src/tests/attribute/attribute_test.cpp
@@ -22,6 +22,7 @@
 #include <vespa/searchlib/util/randomgenerator.h>
 #include <vespa/vespalib/io/fileutil.h>
 #include <vespa/vespalib/testkit/testapp.h>
+#include <vespa/vespalib/util/mmap_file_allocator_factory.h>
 #include <vespa/fastos/file.h>
 #include <cmath>
 #include <iostream>
@@ -275,6 +276,10 @@ private:
     void testPendingCompaction();
     void testConditionalCommit();
 
+    static int64_t stat_size(const vespalib::string& swapfile);
+    int test_paged_attribute(const vespalib::string& name, const vespalib::string& swapfile, const search::attribute::Config& cfg);
+    void test_paged_attributes();
+
 public:
     AttributeTest();
     int Main() override;
@@ -2266,6 +2271,85 @@ AttributeTest::testConditionalCommit() {
     EXPECT_EQUAL(0u, iv.getChangeVectorMemoryUsage().usedBytes());
 }
 
+int64_t
+AttributeTest::stat_size(const vespalib::string& swapfile)
+{
+    auto stat = vespalib::stat(swapfile);
+    ASSERT_TRUE(stat);
+    return stat->_size;
+}
+
+int
+AttributeTest::test_paged_attribute(const vespalib::string& name, const vespalib::string& swapfile, const search::attribute::Config& cfg)
+{
+    int result = 1;
+    LOG(info, "test_paged_attribute '%s'", name.c_str());
+    auto av = createAttribute(name, cfg);
+    auto v = std::dynamic_pointer_cast<IntegerAttribute>(av);
+    ASSERT_TRUE(v);
+    auto size1 = stat_size(swapfile);
+    // Grow mapping from lid to value or multivalue index
+    addClearedDocs(av, 1200);
+    auto size2 = stat_size(swapfile);
+    auto size3 = size2;
+    EXPECT_LESS(size1, size2);
+    if (cfg.collectionType().isMultiValue()) {
+        // Grow multi value mapping
+        for (uint32_t lid = 1; lid < 100; ++lid) {
+            av->clearDoc(lid);
+            for (uint32_t i = 0; i < 50; ++i) {
+                EXPECT_TRUE(v->append(lid, 0, 1));
+            }
+            av->commit();
+        }
+        size3 = stat_size(swapfile);
+        EXPECT_LESS(size2, size3);
+        result += 2;
+    }
+    if (cfg.fastSearch()) {
+        // Grow enum store
+        uint32_t maxlid = cfg.collectionType().isMultiValue() ? 100 : 1200;
+        for (uint32_t lid = 1; lid < maxlid; ++lid) {
+            av->clearDoc(lid);
+            if (cfg.collectionType().isMultiValue()) {
+                for (uint32_t i = 0; i < 50; ++i) {
+                    EXPECT_TRUE(v->append(lid, lid * 100 + i, 1));
+                }
+            } else {
+                EXPECT_TRUE(v->update(lid, lid * 100));
+            }
+            av->commit();
+        }
+        auto size4 = stat_size(swapfile);
+        EXPECT_LESS(size3, size4);
+        result += 4;
+    }
+    return result;
+}
+
+void
+AttributeTest::test_paged_attributes()
+{
+    vespalib::string basedir("mmap-file-allocator-factory-dir");
+    vespalib::alloc::MmapFileAllocatorFactory::instance().setup(basedir);
+    search::attribute::Config cfg1(BasicType::INT32, CollectionType::SINGLE);
+    cfg1.setPaged(true);
+    EXPECT_EQUAL(1, test_paged_attribute("std-int-sv-paged", basedir + "/0.std-int-sv-paged/swapfile", cfg1));
+    search::attribute::Config cfg2(BasicType::INT32, CollectionType::ARRAY);
+    cfg2.setPaged(true);
+    EXPECT_EQUAL(3, test_paged_attribute("std-int-mv-paged", basedir + "/1.std-int-mv-paged/swapfile", cfg2));
+    search::attribute::Config cfg3(BasicType::INT32, CollectionType::SINGLE);
+    cfg3.setPaged(true);
+    cfg3.setFastSearch(true);
+    EXPECT_EQUAL(5, test_paged_attribute("fs-int-sv-paged", basedir + "/2.fs-int-sv-paged/swapfile", cfg3));
+    search::attribute::Config cfg4(BasicType::INT32, CollectionType::ARRAY);
+    cfg4.setPaged(true);
+    cfg4.setFastSearch(true);
+    EXPECT_EQUAL(7, test_paged_attribute("fs-int-mv-paged", basedir + "/3.fs-int-mv-paged/swapfile", cfg4));
+    vespalib::alloc::MmapFileAllocatorFactory::instance().setup("");
+    vespalib::rmdir(basedir, true);
+}
+
 void testNamePrefix() {
     Config cfg(BasicType::INT32, CollectionType::SINGLE);
     AttributeVector::SP vFlat = createAttribute("sfsint32_pc", cfg);
@@ -2348,6 +2432,7 @@ int AttributeTest::Main()
     TEST_DO(testConditionalCommit());
     TEST_DO(testNamePrefix());
     test_multi_value_mapping_has_free_lists_enabled();
+    TEST_DO(test_paged_attributes());
 
     deleteDataDirs();
     TEST_DONE();
diff --git a/searchlib/src/vespa/searchlib/attribute/attributevector.cpp b/searchlib/src/vespa/searchlib/attribute/attributevector.cpp
index 08721b7302e..4c3134d3235 100644
--- a/searchlib/src/vespa/searchlib/attribute/attributevector.cpp
+++ b/searchlib/src/vespa/searchlib/attribute/attributevector.cpp
@@ -836,6 +836,12 @@ AttributeVector::update_config(const Config& cfg)
     drain_hold(1_Mi); // Wait until 1MiB or less on hold
 }
 
+vespalib::alloc::Alloc
+AttributeVector::get_initial_alloc()
+{
+    return (_memory_allocator ? vespalib::alloc::Alloc::alloc_with_allocator(_memory_allocator.get()) : vespalib::alloc::Alloc::alloc());
+}
+
 template bool AttributeVector::append<StringChangeData>(ChangeVectorT< ChangeTemplate<StringChangeData> > &changes, uint32_t , const StringChangeData &, int32_t, bool);
 template bool AttributeVector::update<StringChangeData>(ChangeVectorT< ChangeTemplate<StringChangeData> > &changes, uint32_t , const StringChangeData &);
 template bool AttributeVector::remove<StringChangeData>(ChangeVectorT< ChangeTemplate<StringChangeData> > &changes, uint32_t , const StringChangeData &, int32_t);
diff --git a/searchlib/src/vespa/searchlib/attribute/attributevector.h b/searchlib/src/vespa/searchlib/attribute/attributevector.h
index 2bf2f3a6ed6..8b42b19cc60 100644
--- a/searchlib/src/vespa/searchlib/attribute/attributevector.h
+++ b/searchlib/src/vespa/searchlib/attribute/attributevector.h
@@ -378,6 +378,7 @@ protected:
     virtual void populate_address_space_usage(AddressSpaceUsage& usage) const;
 
     const std::shared_ptr<vespalib::alloc::MemoryAllocator>& get_memory_allocator() const noexcept { return _memory_allocator; }
+    vespalib::alloc::Alloc get_initial_alloc();
 public:
     DECLARE_IDENTIFIABLE_ABSTRACT(AttributeVector);
     bool isLoaded() const { return _loaded; }
diff --git a/searchlib/src/vespa/searchlib/attribute/enumattribute.hpp b/searchlib/src/vespa/searchlib/attribute/enumattribute.hpp
index c0680fd9238..24dfe742120 100644
--- a/searchlib/src/vespa/searchlib/attribute/enumattribute.hpp
+++ b/searchlib/src/vespa/searchlib/attribute/enumattribute.hpp
@@ -14,7 +14,7 @@ EnumAttribute<B>::
 EnumAttribute(const vespalib::string &baseFileName,
               const AttributeVector::Config &cfg)
     : B(baseFileName, cfg),
-      _enumStore(cfg.fastSearch(), cfg.get_dictionary_config())
+      _enumStore(cfg.fastSearch(), cfg.get_dictionary_config(), this->get_memory_allocator())
 {
     this->setEnum(true);
 }
diff --git a/searchlib/src/vespa/searchlib/attribute/multivalueattribute.hpp b/searchlib/src/vespa/searchlib/attribute/multivalueattribute.hpp
index c0524a4d043..326dbd4a380 100644
--- a/searchlib/src/vespa/searchlib/attribute/multivalueattribute.hpp
+++ b/searchlib/src/vespa/searchlib/attribute/multivalueattribute.hpp
@@ -28,7 +28,7 @@ MultiValueAttribute(const vespalib::string &baseFileName,
                                                                8 * 1024,
                                                                cfg.getGrowStrategy().getMultiValueAllocGrowFactor(),
                                                                multivalueattribute::enable_free_lists),
-                 cfg.getGrowStrategy().to_generic_strategy(), {})
+                 cfg.getGrowStrategy().to_generic_strategy(), this->get_memory_allocator())
 {
 }
 
diff --git a/searchlib/src/vespa/searchlib/attribute/singleenumattribute.cpp b/searchlib/src/vespa/searchlib/attribute/singleenumattribute.cpp
index d049e53c0a2..cffd52f3dc4 100644
--- a/searchlib/src/vespa/searchlib/attribute/singleenumattribute.cpp
+++ b/searchlib/src/vespa/searchlib/attribute/singleenumattribute.cpp
@@ -14,11 +14,11 @@ namespace search {
 using attribute::Config;
 
 SingleValueEnumAttributeBase::
-SingleValueEnumAttributeBase(const Config & c, GenerationHolder &genHolder)
+SingleValueEnumAttributeBase(const Config & c, GenerationHolder &genHolder, const vespalib::alloc::Alloc& initial_alloc)
     : _enumIndices(c.getGrowStrategy().getDocsInitialCapacity(),
                    c.getGrowStrategy().getDocsGrowPercent(),
                    c.getGrowStrategy().getDocsGrowDelta(),
-                   genHolder)
+                   genHolder, initial_alloc)
 {
 }
 
diff --git a/searchlib/src/vespa/searchlib/attribute/singleenumattribute.h b/searchlib/src/vespa/searchlib/attribute/singleenumattribute.h
index a6bcfd92449..8fcef670fb0 100644
--- a/searchlib/src/vespa/searchlib/attribute/singleenumattribute.h
+++ b/searchlib/src/vespa/searchlib/attribute/singleenumattribute.h
@@ -30,7 +30,7 @@ public:
     IEnumStore::Index getEnumIndex(DocId docId) const { return _enumIndices[docId]; }
     EnumHandle getE(DocId doc) const { return _enumIndices[doc].ref(); }
 protected:
-    SingleValueEnumAttributeBase(const attribute::Config & c, GenerationHolder &genHolder);
+    SingleValueEnumAttributeBase(const attribute::Config & c, GenerationHolder &genHolder, const vespalib::alloc::Alloc& initial_alloc);
     ~SingleValueEnumAttributeBase();
     AttributeVector::DocId addDoc(bool & incGeneration);
 
diff --git a/searchlib/src/vespa/searchlib/attribute/singleenumattribute.hpp b/searchlib/src/vespa/searchlib/attribute/singleenumattribute.hpp
index a51c9804cf2..56aa6672696 100644
--- a/searchlib/src/vespa/searchlib/attribute/singleenumattribute.hpp
+++ b/searchlib/src/vespa/searchlib/attribute/singleenumattribute.hpp
@@ -16,7 +16,7 @@ SingleValueEnumAttribute<B>::
 SingleValueEnumAttribute(const vespalib::string &baseFileName,
                          const AttributeVector::Config &cfg)
     : B(baseFileName, cfg),
-      SingleValueEnumAttributeBase(cfg, getGenerationHolder())
+      SingleValueEnumAttributeBase(cfg, getGenerationHolder(), this->get_initial_alloc())
 {
 }
 
diff --git a/searchlib/src/vespa/searchlib/attribute/singlenumericattribute.hpp b/searchlib/src/vespa/searchlib/attribute/singlenumericattribute.hpp
index 89c2a21231b..ef618262bcd 100644
--- a/searchlib/src/vespa/searchlib/attribute/singlenumericattribute.hpp
+++ b/searchlib/src/vespa/searchlib/attribute/singlenumericattribute.hpp
@@ -19,7 +19,8 @@ SingleValueNumericAttribute(const vespalib::string & baseFileName, const Attribu
     _data(c.getGrowStrategy().getDocsInitialCapacity(),
           c.getGrowStrategy().getDocsGrowPercent(),
           c.getGrowStrategy().getDocsGrowDelta(),
-          getGenerationHolder())
+          getGenerationHolder(),
+          this->get_initial_alloc())
 { }
 
 template <typename B>