Merge pull request #18723 from vespa-engine/balder/add-swapable-attribute-option

Balder/add swapable attribute option

9 files changed, 97 insertions, 97 deletions

diff --git a/searchlib/src/tests/attribute/attributemanager/attributemanager_test.cpp b/searchlib/src/tests/attribute/attributemanager/attributemanager_test.cpp
index 721e7e1c805..877b5b9e427 100644
--- a/searchlib/src/tests/attribute/attributemanager/attributemanager_test.cpp
+++ b/searchlib/src/tests/attribute/attributemanager/attributemanager_test.cpp
@@ -20,43 +20,13 @@ using namespace search::attribute;
 using std::shared_ptr;
 using vespalib::stringref;
 
-typedef BasicType      BT;
-typedef CollectionType CT;
-typedef AttributeVector::SP             AVSP;
+using BT = BasicType;
+using CT = CollectionType;
+using AVSP = AttributeVector::SP;
 
 namespace search {
 
-class AttributeManagerTest : public vespalib::TestApp
-{
-private:
-    void verifyLoad(AttributeVector & v);
-    void testLoad();
-    void testGuards();
-    void testConfigConvert();
-    void testContext();
-    void can_get_readable_attribute_vector_by_name();
-
-    bool
-    assertDataType(BT::Type exp,
-                   AttributesConfig::Attribute::Datatype in);
-
-    bool
-    assertCollectionType(CollectionType exp,
-                         AttributesConfig::Attribute::Collectiontype in,
-                         bool removeIfZ = false,
-                         bool createIfNe = false);
-
-public:
-    AttributeManagerTest()
-    {
-    }
-    int Main() override;
-};
-
-
-typedef MultiValueNumericAttribute< IntegerAttributeTemplate<int32_t>,
-                                    multivalue::Value<int32_t> >
-TestAttributeBase;
+using TestAttributeBase = MultiValueNumericAttribute< IntegerAttributeTemplate<int32_t>, multivalue::Value<int32_t> >;
 
 class TestAttribute : public TestAttributeBase
 {
@@ -73,8 +43,7 @@ public:
 };
 
 
-void
-AttributeManagerTest::testGuards()
+TEST("Test attribute guards")
 {
     AttributeVector::SP vec(new TestAttribute("mvint") );
     TestAttribute * v = static_cast<TestAttribute *> (vec.get());
@@ -134,7 +103,7 @@ AttributeManagerTest::testGuards()
 
 
 void
-AttributeManagerTest::verifyLoad(AttributeVector & v)
+verifyLoad(AttributeVector & v)
 {
     EXPECT_TRUE( !v.isLoaded() );
     EXPECT_TRUE( v.load() );
@@ -143,8 +112,7 @@ AttributeManagerTest::verifyLoad(AttributeVector & v)
 }
 
 
-void
-AttributeManagerTest::testLoad()
+TEST("Test loading of attributes")
 {
     {
         TestAttributeBase v("mvint");
@@ -172,14 +140,14 @@ AttributeManagerTest::testLoad()
         verifyLoad(v);
     }
     {
-        AttributeVector::Config config(BT::INT32,
+        Config config(BT::INT32,
                                        CollectionType::ARRAY);
         TestAttributeBase v("mvint", config);
         verifyLoad(v);
     }
     {
         AttributeManager manager;
-        AttributeVector::Config config(BT::INT32,
+        Config config(BT::INT32,
                                        CollectionType::ARRAY);
         EXPECT_TRUE(manager.addVector("mvint", config));
         AttributeManager::AttributeList list;
@@ -193,7 +161,7 @@ AttributeManagerTest::testLoad()
 
 
 bool
-AttributeManagerTest::assertDataType(BT::Type exp, AttributesConfig::Attribute::Datatype in)
+assertDataType(BT::Type exp, AttributesConfig::Attribute::Datatype in)
 {
     AttributesConfig::Attribute a;
     a.datatype = in;
@@ -202,25 +170,22 @@ AttributeManagerTest::assertDataType(BT::Type exp, AttributesConfig::Attribute::
 
 
 bool
-AttributeManagerTest::
 assertCollectionType(CollectionType exp, AttributesConfig::Attribute::Collectiontype in,
-                     bool removeIfZ, bool createIfNe)
+                     bool removeIfZ = false, bool createIfNe = false)
 {
     AttributesConfig::Attribute a;
     a.collectiontype = in;
     a.removeifzero = removeIfZ;
     a.createifnonexistent = createIfNe;
-    AttributeVector::Config out = ConfigConverter::convert(a);
+    Config out = ConfigConverter::convert(a);
     return EXPECT_EQUAL(exp.type(), out.collectionType().type()) &&
         EXPECT_EQUAL(exp.removeIfZero(), out.collectionType().removeIfZero()) &&
         EXPECT_EQUAL(exp.createIfNonExistant(), out.collectionType().createIfNonExistant());
 }
 
 
-void
-AttributeManagerTest::testConfigConvert()
+TEST("require that config can be converted")
 {
-    // typedef AttributeVector::Config AVC;
     typedef BT AVBT;
     typedef CollectionType AVCT;
     using CACA = AttributesConfig::Attribute;
@@ -271,11 +236,17 @@ AttributeManagerTest::testConfigConvert()
         a.ismutable = true;
         EXPECT_TRUE(CC::convert(a).isMutable());
     }
+    {
+        CACA a;
+        EXPECT_TRUE(!CC::convert(a).paged());
+        a.paged = true;
+        EXPECT_TRUE(CC::convert(a).paged());
+    }
     { // tensor
         CACA a;
         a.datatype = CACAD::TENSOR;
         a.tensortype = "tensor(x[5])";
-        AttributeVector::Config out = ConfigConverter::convert(a);
+        Config out = ConfigConverter::convert(a);
         EXPECT_EQUAL("tensor(x[5])", out.tensorType().to_spec());
     }
     { // distance metric (default)
@@ -334,8 +305,7 @@ bool gt_attribute(const attribute::IAttributeVector * a, const attribute::IAttri
     return a->getName() < b->getName();
 }
 
-void
-AttributeManagerTest::testContext()
+TEST("test the attribute context")
 {
     std::vector<AVSP> attrs;
     // create various attributes vectors
@@ -370,13 +340,13 @@ AttributeManagerTest::testContext()
         }
 
         for (uint32_t i = 0; i < 2; ++i) {
-            EXPECT_TRUE(first->getAttribute("sint32") != NULL);
-            EXPECT_TRUE(first->getAttribute("aint32") != NULL);
-            EXPECT_TRUE(first->getAttribute("wsint32") != NULL);
-            EXPECT_TRUE(first->getAttributeStableEnum("wsint32") != NULL);
+            EXPECT_TRUE(first->getAttribute("sint32") != nullptr);
+            EXPECT_TRUE(first->getAttribute("aint32") != nullptr);
+            EXPECT_TRUE(first->getAttribute("wsint32") != nullptr);
+            EXPECT_TRUE(first->getAttributeStableEnum("wsint32") != nullptr);
         }
-        EXPECT_TRUE(first->getAttribute("foo") == NULL);
-        EXPECT_TRUE(first->getAttribute("bar") == NULL);
+        EXPECT_TRUE(first->getAttribute("foo") == nullptr);
+        EXPECT_TRUE(first->getAttribute("bar") == nullptr);
 
         // one generation guard taken per attribute asked for
         for (uint32_t i = 0; i < attrs.size(); ++i) {
@@ -388,10 +358,10 @@ AttributeManagerTest::testContext()
         {
             IAttributeContext::UP second = manager.createContext();
 
-            EXPECT_TRUE(second->getAttribute("sint32") != NULL);
-            EXPECT_TRUE(second->getAttribute("aint32") != NULL);
-            EXPECT_TRUE(second->getAttribute("wsint32") != NULL);
-            EXPECT_TRUE(second->getAttributeStableEnum("wsint32") != NULL);
+            EXPECT_TRUE(second->getAttribute("sint32") != nullptr);
+            EXPECT_TRUE(second->getAttribute("aint32") != nullptr);
+            EXPECT_TRUE(second->getAttribute("wsint32") != nullptr);
+            EXPECT_TRUE(second->getAttributeStableEnum("wsint32") != nullptr);
 
             // two generation guards taken per attribute asked for
             for (uint32_t i = 0; i < attrs.size(); ++i) {
@@ -428,8 +398,7 @@ AttributeManagerTest::testContext()
     }
 }
 
-void
-AttributeManagerTest::can_get_readable_attribute_vector_by_name()
+TEST("require that we can get readable attribute by name")
 {
     auto attr = AttributeFactory::createAttribute("cool_attr", Config(BT::INT32, CT::SINGLE));
     // Ensure there's something to actually load, or fetching the attribute will throw.
@@ -443,20 +412,7 @@ AttributeManagerTest::can_get_readable_attribute_vector_by_name()
     EXPECT_TRUE(av.get() == nullptr);
 }
 
-int AttributeManagerTest::Main()
-{
-    TEST_INIT("attributemanager_test");
-
-    testLoad();
-    testGuards();
-    testConfigConvert();
-    testContext();
-    can_get_readable_attribute_vector_by_name();
-
-    TEST_DONE();
-}
-
 } // namespace search
 
 
-TEST_APPHOOK(search::AttributeManagerTest);
+TEST_MAIN() { TEST_RUN_ALL(); }
diff --git a/searchlib/src/tests/attribute/tensorattribute/tensorattribute_test.cpp b/searchlib/src/tests/attribute/tensorattribute/tensorattribute_test.cpp
index b27c26d1139..605fcb538d6 100644
--- a/searchlib/src/tests/attribute/tensorattribute/tensorattribute_test.cpp
+++ b/searchlib/src/tests/attribute/tensorattribute/tensorattribute_test.cpp
@@ -26,6 +26,7 @@
 #include <vespa/vespalib/io/fileutil.h>
 #include <vespa/vespalib/test/insertion_operators.h>
 #include <vespa/vespalib/testkit/test_kit.h>
+#include <vespa/vespalib/util/mmap_file_allocator_factory.h>
 #include <vespa/searchlib/util/bufferwriter.h>
 
 #include <vespa/log/log.h>
@@ -256,6 +257,7 @@ struct FixtureTraits {
     bool use_direct_tensor_attribute = false;
     bool enable_hnsw_index = false;
     bool use_mock_index = false;
+    bool use_mmap_file_allocator = false;
 
     FixtureTraits dense() && {
         use_dense_tensor_attribute = true;
@@ -263,6 +265,11 @@ struct FixtureTraits {
         return *this;
     }
 
+    FixtureTraits mmap_file_allocator() && {
+        use_mmap_file_allocator = true;
+        return *this;
+    }
+
     FixtureTraits hnsw() && {
         use_dense_tensor_attribute = true;
         enable_hnsw_index = true;
@@ -327,6 +334,9 @@ struct Fixture {
         if (_cfg.tensorType().is_dense()) {
             _denseTensors = true;
         }
+        if (_traits.use_mmap_file_allocator) {
+            _cfg.setPaged(true);
+        }
         if (_traits.use_mock_index) {
             _index_factory = std::make_unique<MockNearestNeighborIndexFactory>();
         } else {
@@ -1000,4 +1010,17 @@ TEST_F("NN blueprint handles strong filter triggering brute force search", Neare
     EXPECT_FALSE(bp->may_approximate());
 }
 
+TEST("Dense tensor attribute with paged flag uses mmap file allocator")
+{
+    vespalib::string basedir("mmap-file-allocator-factory-dir");
+    vespalib::alloc::MmapFileAllocatorFactory::instance().setup(basedir);
+    {
+        Fixture f(vec_2d_spec, FixtureTraits().dense().mmap_file_allocator());
+        vespalib::string allocator_dir(basedir + "/0.my_attr");
+        EXPECT_TRUE(vespalib::isDirectory(allocator_dir));
+    }
+    vespalib::alloc::MmapFileAllocatorFactory::instance().setup("");
+    vespalib::rmdir(basedir, true);
+}
+
 TEST_MAIN() { TEST_RUN_ALL(); }
diff --git a/searchlib/src/tests/tensor/dense_tensor_store/dense_tensor_store_test.cpp b/searchlib/src/tests/tensor/dense_tensor_store/dense_tensor_store_test.cpp
index eb8d89edc07..75e231a815c 100644
--- a/searchlib/src/tests/tensor/dense_tensor_store/dense_tensor_store_test.cpp
+++ b/searchlib/src/tests/tensor/dense_tensor_store/dense_tensor_store_test.cpp
@@ -1,5 +1,8 @@
 // Copyright 2017 Yahoo Holdings. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+#include <vespa/log/log.h>
+LOG_SETUP("dense_tensor_store_test");
 #include <vespa/vespalib/testkit/test_kit.h>
+#include <vespa/vespalib/util/memory_allocator.h>
 #include <vespa/searchlib/tensor/dense_tensor_store.h>
 #include <vespa/eval/eval/simple_value.h>
 #include <vespa/eval/eval/tensor_spec.h>
@@ -7,9 +10,6 @@
 #include <vespa/eval/eval/value_type.h>
 #include <vespa/eval/eval/test/value_compare.h>
 
-#include <vespa/log/log.h>
-LOG_SETUP("dense_tensor_store_test");
-
 using search::tensor::DenseTensorStore;
 using vespalib::eval::SimpleValue;
 using vespalib::eval::TensorSpec;
@@ -28,7 +28,7 @@ struct Fixture
 {
     DenseTensorStore store;
     Fixture(const vespalib::string &tensorType)
-        : store(ValueType::from_spec(tensorType))
+        : store(ValueType::from_spec(tensorType), {})
     {}
     void assertSetAndGetTensor(const TensorSpec &tensorSpec) {
         Value::UP expTensor = makeTensor(tensorSpec);
diff --git a/searchlib/src/vespa/searchlib/attribute/attributevector.h b/searchlib/src/vespa/searchlib/attribute/attributevector.h
index 3c27f00a022..99b0c923c90 100644
--- a/searchlib/src/vespa/searchlib/attribute/attributevector.h
+++ b/searchlib/src/vespa/searchlib/attribute/attributevector.h
@@ -282,6 +282,14 @@ public:
 
     virtual IExtendAttribute * getExtendInterface();
 
+    /**
+     * Returns the number of readers holding a generation guard.
+     * Should be called by the writer thread.
+     **/
+    uint32_t getGenerationRefCount(generation_t gen) const {
+        return _genHandler.getGenerationRefCount(gen);
+    }
+
 protected:
     /**
      * Called when a new document has been added, but only for
@@ -291,14 +299,6 @@ protected:
      **/
     virtual bool onAddDoc(DocId) { return false; }
 
-    /**
-     * Returns the number of readers holding a generation guard.
-     * Should be called by the writer thread.
-     */
-    uint32_t getGenerationRefCount(generation_t gen) const {
-        return _genHandler.getGenerationRefCount(gen);
-    }
-
     const GenerationHandler & getGenerationHandler() const {
         return _genHandler;
     }
diff --git a/searchlib/src/vespa/searchlib/attribute/configconverter.cpp b/searchlib/src/vespa/searchlib/attribute/configconverter.cpp
index 6387edc588b..625502ee32b 100644
--- a/searchlib/src/vespa/searchlib/attribute/configconverter.cpp
+++ b/searchlib/src/vespa/searchlib/attribute/configconverter.cpp
@@ -108,6 +108,7 @@ ConfigConverter::convert(const AttributesConfig::Attribute & cfg)
     retval.setIsFilter(cfg.enableonlybitvector);
     retval.setFastAccess(cfg.fastaccess);
     retval.setMutable(cfg.ismutable);
+    retval.setPaged(cfg.paged);
     predicateParams.setArity(cfg.arity);
     predicateParams.setBounds(cfg.lowerbound, cfg.upperbound);
     predicateParams.setDensePostingListThreshold(cfg.densepostinglistthreshold);
diff --git a/searchlib/src/vespa/searchlib/tensor/dense_tensor_attribute.cpp b/searchlib/src/vespa/searchlib/tensor/dense_tensor_attribute.cpp
index be15967ab10..d0943b372be 100644
--- a/searchlib/src/vespa/searchlib/tensor/dense_tensor_attribute.cpp
+++ b/searchlib/src/vespa/searchlib/tensor/dense_tensor_attribute.cpp
@@ -10,6 +10,8 @@
 #include <vespa/searchlib/attribute/load_utils.h>
 #include <vespa/searchlib/attribute/readerbase.h>
 #include <vespa/vespalib/data/slime/inserter.h>
+#include <vespa/vespalib/util/memory_allocator.h>
+#include <vespa/vespalib/util/mmap_file_allocator_factory.h>
 
 #include <vespa/log/log.h>
 LOG_SETUP(".searchlib.tensor.dense_tensor_attribute");
@@ -72,6 +74,15 @@ can_use_index_save_file(const search::attribute::Config &config, const search::a
     return true;
 }
 
+std::unique_ptr<vespalib::alloc::MemoryAllocator>
+make_memory_allocator(const vespalib::string& name, bool swappable)
+{
+    if (swappable) {
+        return vespalib::alloc::MmapFileAllocatorFactory::instance().make_memory_allocator(name);
+    }
+    return {};
+}
+
 }
 
 void
@@ -114,7 +125,7 @@ DenseTensorAttribute::memory_usage() const
 DenseTensorAttribute::DenseTensorAttribute(vespalib::stringref baseFileName, const Config& cfg,
                                            const NearestNeighborIndexFactory& index_factory)
     : TensorAttribute(baseFileName, cfg, _denseTensorStore),
-      _denseTensorStore(cfg.tensorType()),
+      _denseTensorStore(cfg.tensorType(), make_memory_allocator(getName(), cfg.paged())),
       _index()
 {
     if (cfg.hnsw_index_params().has_value()) {
diff --git a/searchlib/src/vespa/searchlib/tensor/dense_tensor_attribute.h b/searchlib/src/vespa/searchlib/tensor/dense_tensor_attribute.h
index 0884b776ca5..7eb8f2510f7 100644
--- a/searchlib/src/vespa/searchlib/tensor/dense_tensor_attribute.h
+++ b/searchlib/src/vespa/searchlib/tensor/dense_tensor_attribute.h
@@ -29,7 +29,7 @@ private:
 public:
     DenseTensorAttribute(vespalib::stringref baseFileName, const Config& cfg,
                          const NearestNeighborIndexFactory& index_factory = DefaultNearestNeighborIndexFactory());
-    virtual ~DenseTensorAttribute();
+    ~DenseTensorAttribute() override;
     // Implements AttributeVector and ITensorAttribute
     uint32_t clearDoc(DocId docId) override;
     void setTensor(DocId docId, const vespalib::eval::Value &tensor) override;
diff --git a/searchlib/src/vespa/searchlib/tensor/dense_tensor_store.cpp b/searchlib/src/vespa/searchlib/tensor/dense_tensor_store.cpp
index e6da839da2e..13796d35dec 100644
--- a/searchlib/src/vespa/searchlib/tensor/dense_tensor_store.cpp
+++ b/searchlib/src/vespa/searchlib/tensor/dense_tensor_store.cpp
@@ -40,8 +40,9 @@ DenseTensorStore::TensorSizeCalc::alignedSize() const
     return my_align(bufSize(), DENSE_TENSOR_ALIGNMENT);
 }
 
-DenseTensorStore::BufferType::BufferType(const TensorSizeCalc &tensorSizeCalc)
-    : vespalib::datastore::BufferType<char>(tensorSizeCalc.alignedSize(), MIN_BUFFER_ARRAYS, RefType::offsetSize())
+DenseTensorStore::BufferType::BufferType(const TensorSizeCalc &tensorSizeCalc, std::unique_ptr<vespalib::alloc::MemoryAllocator> allocator)
+    : vespalib::datastore::BufferType<char>(tensorSizeCalc.alignedSize(), MIN_BUFFER_ARRAYS, RefType::offsetSize()),
+      _allocator(std::move(allocator))
 {}
 
 DenseTensorStore::BufferType::~BufferType() = default;
@@ -53,11 +54,17 @@ DenseTensorStore::BufferType::cleanHold(void *buffer, size_t offset,
     memset(static_cast<char *>(buffer) + offset, 0, numElems);
 }
 
-DenseTensorStore::DenseTensorStore(const ValueType &type)
+const vespalib::alloc::MemoryAllocator*
+DenseTensorStore::BufferType::get_memory_allocator() const
+{
+    return _allocator.get();
+}
+
+DenseTensorStore::DenseTensorStore(const ValueType &type, std::unique_ptr<vespalib::alloc::MemoryAllocator> allocator)
     : TensorStore(_concreteStore),
       _concreteStore(),
       _tensorSizeCalc(type),
-      _bufferType(_tensorSizeCalc),
+      _bufferType(_tensorSizeCalc, std::move(allocator)),
       _type(type),
       _emptySpace()
 {
diff --git a/searchlib/src/vespa/searchlib/tensor/dense_tensor_store.h b/searchlib/src/vespa/searchlib/tensor/dense_tensor_store.h
index 638f602d613..dad28642e67 100644
--- a/searchlib/src/vespa/searchlib/tensor/dense_tensor_store.h
+++ b/searchlib/src/vespa/searchlib/tensor/dense_tensor_store.h
@@ -36,10 +36,12 @@ public:
     class BufferType : public vespalib::datastore::BufferType<char>
     {
         using CleanContext = vespalib::datastore::BufferType<char>::CleanContext;
+        std::unique_ptr<vespalib::alloc::MemoryAllocator> _allocator;
     public:
-        BufferType(const TensorSizeCalc &tensorSizeCalc);
+        BufferType(const TensorSizeCalc &tensorSizeCalc, std::unique_ptr<vespalib::alloc::MemoryAllocator> allocator);
         ~BufferType() override;
         void cleanHold(void *buffer, size_t offset, ElemCount numElems, CleanContext cleanCtx) override;
+        const vespalib::alloc::MemoryAllocator* get_memory_allocator() const override;
     };
 private:
     DataStoreType _concreteStore;
@@ -55,7 +57,7 @@ private:
     setDenseTensor(const TensorType &tensor);
 
 public:
-    DenseTensorStore(const ValueType &type);
+    DenseTensorStore(const ValueType &type, std::unique_ptr<vespalib::alloc::MemoryAllocator> allocator);
     ~DenseTensorStore() override;
 
     const ValueType &type() const { return _type; }