SERVER-73785 Improve SBE ArrayEnumerator to avoid repeated strlen calls

10 files changed, 60 insertions, 39 deletions

diff --git a/src/mongo/db/exec/sbe/sbe_mkobj_test.cpp b/src/mongo/db/exec/sbe/sbe_mkobj_test.cpp
index 40df6e6491c..64433238c9c 100644
--- a/src/mongo/db/exec/sbe/sbe_mkobj_test.cpp
+++ b/src/mongo/db/exec/sbe/sbe_mkobj_test.cpp
@@ -50,7 +50,7 @@ public:
         be += 4;
         const char* end = be + obj.objsize();
         while (*be != 0) {
-            auto sv = bson::fieldNameView(be);
+            auto sv = bson::fieldNameAndLength(be);
             auto [tag, val] = bson::convertFrom<false>(be, end, sv.size());
             be = bson::advance(be, sv.size());
 
diff --git a/src/mongo/db/exec/sbe/stages/makeobj.cpp b/src/mongo/db/exec/sbe/stages/makeobj.cpp
index a2cb193240b..9f14c32eef5 100644
--- a/src/mongo/db/exec/sbe/stages/makeobj.cpp
+++ b/src/mongo/db/exec/sbe/stages/makeobj.cpp
@@ -191,7 +191,7 @@ void MakeObjStageBase<MakeObjOutputType::object>::produceObject() {
                 // Skip document length.
                 be += 4;
                 while (*be != 0) {
-                    auto sv = bson::fieldNameView(be);
+                    auto sv = bson::fieldNameAndLength(be);
                     auto key = StringMapHasher{}.hashed_key(StringData(sv));
 
                     if (!isFieldProjectedOrRestricted(key)) {
diff --git a/src/mongo/db/exec/sbe/stages/scan.cpp b/src/mongo/db/exec/sbe/stages/scan.cpp
index 6ff452ed30e..d0f19ed8cdf 100644
--- a/src/mongo/db/exec/sbe/stages/scan.cpp
+++ b/src/mongo/db/exec/sbe/stages/scan.cpp
@@ -474,7 +474,7 @@ PlanState ScanStage::getNext() {
             auto name = StringData{_fields[0]};
             auto [tag, val] = [start, last, end, name] {
                 for (auto bsonElement = start; bsonElement != last;) {
-                    auto field = bson::fieldNameView(bsonElement);
+                    auto field = bson::fieldNameAndLength(bsonElement);
                     if (field == name) {
                         return bson::convertFrom<true>(bsonElement, end, field.size());
                     }
@@ -499,7 +499,7 @@ PlanState ScanStage::getNext() {
                 // machine instructions) in front of the hashtable. When we "miss" in the bloom
                 // filter, we can quickly skip over a field without having to generate the hash for
                 // the field.
-                auto field = bson::fieldNameView(bsonElement);
+                auto field = bson::fieldNameAndLength(bsonElement);
                 const size_t offset = computeFieldMaskOffset(field.rawData(), field.size());
                 if (!(_fieldsBloomFilter & computeFieldMask(offset))) {
                     bsonElement = bson::advance(bsonElement, field.size());
@@ -1032,7 +1032,7 @@ PlanState ParallelScanStage::getNext() {
             accessor->reset();
         }
         while (*be != 0) {
-            auto sv = bson::fieldNameView(be);
+            auto sv = bson::fieldNameAndLength(be);
             if (auto it = _fieldAccessors.find(sv); it != _fieldAccessors.end()) {
                 // Found the field so convert it to Value.
                 auto [tag, val] = bson::convertFrom<true>(be, end, sv.size());
diff --git a/src/mongo/db/exec/sbe/values/bson.cpp b/src/mongo/db/exec/sbe/values/bson.cpp
index 446a71b2301..877fd6405b5 100644
--- a/src/mongo/db/exec/sbe/values/bson.cpp
+++ b/src/mongo/db/exec/sbe/values/bson.cpp
@@ -167,13 +167,13 @@ std::pair<value::TypeTags, value::Value> convertFrom(const char* be,
             if constexpr (View) {
                 return {value::TypeTags::bsonObject, value::bitcastFrom<const char*>(be)};
             }
-            // Skip document length.
+            const auto objEnd = be + ConstDataView(be).read<LittleEndian<uint32_t>>();
             be += 4;
             auto [tag, val] = value::makeNewObject();
-            auto obj = value::getObjectView(val);
+            const auto obj = value::getObjectView(val);
 
-            while (*be != 0) {
-                auto sv = bson::fieldNameView(be);
+            while (be != objEnd - 1) {
+                auto sv = bson::fieldNameAndLength(be);
 
                 auto [tag, val] = convertFrom<false>(be, end, sv.size());
                 obj->push_back(sv, tag, val);
@@ -187,12 +187,13 @@ std::pair<value::TypeTags, value::Value> convertFrom(const char* be,
                 return {value::TypeTags::bsonArray, value::bitcastFrom<const char*>(be)};
             }
             // Skip array length.
+            const auto arrEnd = be + ConstDataView(be).read<LittleEndian<uint32_t>>();
             be += 4;
             auto [tag, val] = value::makeNewArray();
             auto arr = value::getArrayView(val);
 
-            while (*be != 0) {
-                auto sv = bson::fieldNameView(be);
+            while (be != arrEnd - 1) {
+                auto sv = bson::fieldNameAndLength(be);
 
                 auto [tag, val] = convertFrom<false>(be, end, sv.size());
                 arr->push_back(tag, val);
diff --git a/src/mongo/db/exec/sbe/values/bson.h b/src/mongo/db/exec/sbe/values/bson.h
index 0965ba11421..622c9c1da98 100644
--- a/src/mongo/db/exec/sbe/values/bson.h
+++ b/src/mongo/db/exec/sbe/values/bson.h
@@ -48,10 +48,14 @@ std::pair<value::TypeTags, value::Value> convertFrom(const BSONElement& elem) {
 
 const char* advance(const char* be, size_t fieldNameSize);
 
-inline auto fieldNameView(const char* be) noexcept {
+inline auto fieldNameAndLength(const char* be) noexcept {
     return StringData{be + 1};
 }
 
+inline const char* fieldNameRaw(const char* be) noexcept {
+    return be + 1;
+}
+
 template <class ArrayBuilder>
 void convertToBsonObj(ArrayBuilder& builder, value::Array* arr);
 
diff --git a/src/mongo/db/exec/sbe/values/makeobj_spec.h b/src/mongo/db/exec/sbe/values/makeobj_spec.h
index 9e6ee5a9119..f260e94d812 100644
--- a/src/mongo/db/exec/sbe/values/makeobj_spec.h
+++ b/src/mongo/db/exec/sbe/values/makeobj_spec.h
@@ -111,10 +111,11 @@ public:
         if (rootTag == value::TypeTags::bsonObject) {
             if (!(nFieldsNeededIfInclusion == 0 && fieldBehavior == FieldBehavior::keep)) {
                 auto be = value::bitcastTo<const char*>(rootVal);
+                const auto end = be + ConstDataView(be).read<LittleEndian<uint32_t>>();
                 // Skip document length.
                 be += 4;
-                while (*be != 0) {
-                    auto sv = bson::fieldNameView(be);
+                while (be != end - 1) {
+                    auto sv = bson::fieldNameAndLength(be);
                     auto key = StringMapHasher{}.hashed_key(StringData(sv));
                     auto nextBe = bson::advance(be, sv.size());
 
diff --git a/src/mongo/db/exec/sbe/values/value.cpp b/src/mongo/db/exec/sbe/values/value.cpp
index a24883a7be9..69252625fdd 100644
--- a/src/mongo/db/exec/sbe/values/value.cpp
+++ b/src/mongo/db/exec/sbe/values/value.cpp
@@ -859,8 +859,7 @@ std::pair<TypeTags, Value> ArrayEnumerator::getViewOfValue() const {
     } else if (_arraySet) {
         return {_iter->first, _iter->second};
     } else {
-        auto sv = bson::fieldNameView(_arrayCurrent);
-        return bson::convertFrom<true>(_arrayCurrent, _arrayEnd, sv.size());
+        return bson::convertFrom<true>(_arrayCurrent, _arrayEnd, _fieldNameSize);
     }
 }
 
@@ -878,12 +877,24 @@ bool ArrayEnumerator::advance() {
 
         return _iter != _arraySet->values().end();
     } else {
-        if (*_arrayCurrent != 0) {
-            auto sv = bson::fieldNameView(_arrayCurrent);
-            _arrayCurrent = bson::advance(_arrayCurrent, sv.size());
+        if (_arrayCurrent != _arrayEnd - 1) {
+            _arrayCurrent = bson::advance(_arrayCurrent, _fieldNameSize);
+
+            // Hand rolled strlen() code is faster than calling into shared library
+            // strlen() for small strings. Since this is a BSON array we are guaranteed
+            // the key names are small.
+            if (_arrayCurrent != _arrayEnd - 1) {
+                size_t keySize = 0;
+                auto* fieldName = bson::fieldNameRaw(_arrayCurrent);
+                while (*fieldName) {
+                    keySize++;
+                    fieldName++;
+                }
+                _fieldNameSize = keySize;
+            }
         }
 
-        return *_arrayCurrent != 0;
+        return _arrayCurrent != _arrayEnd - 1;
     }
 }
 
@@ -891,7 +902,7 @@ std::pair<TypeTags, Value> ObjectEnumerator::getViewOfValue() const {
     if (_object) {
         return _object->getAt(_index);
     } else {
-        auto sv = bson::fieldNameView(_objectCurrent);
+        auto sv = bson::fieldNameAndLength(_objectCurrent);
         return bson::convertFrom<true>(_objectCurrent, _objectEnd, sv.size());
     }
 }
@@ -905,7 +916,7 @@ bool ObjectEnumerator::advance() {
         return _index < _object->size();
     } else {
         if (*_objectCurrent != 0) {
-            auto sv = bson::fieldNameView(_objectCurrent);
+            auto sv = bson::fieldNameAndLength(_objectCurrent);
             _objectCurrent = bson::advance(_objectCurrent, sv.size());
         }
 
@@ -923,7 +934,7 @@ StringData ObjectEnumerator::getFieldName() const {
         }
     } else {
         if (*_objectCurrent != 0) {
-            return bson::fieldNameView(_objectCurrent);
+            return bson::fieldNameAndLength(_objectCurrent);
         } else {
             return ""_sd;
         }
diff --git a/src/mongo/db/exec/sbe/values/value.h b/src/mongo/db/exec/sbe/values/value.h
index afd11b94662..c6c88561aa4 100644
--- a/src/mongo/db/exec/sbe/values/value.h
+++ b/src/mongo/db/exec/sbe/values/value.h
@@ -1643,6 +1643,9 @@ public:
                 auto bson = getRawPointerView(val);
                 _arrayCurrent = bson + 4;
                 _arrayEnd = bson + ConstDataView(bson).read<LittleEndian<uint32_t>>();
+                if (_arrayCurrent != _arrayEnd - 1) {
+                    _fieldNameSize = strlen(_arrayCurrent + 1);
+                }
             } else {
                 MONGO_UNREACHABLE;
             }
@@ -1661,7 +1664,7 @@ public:
         } else if (_arraySet) {
             return _iter == _arraySet->values().end();
         } else {
-            return *_arrayCurrent == 0;
+            return _arrayCurrent == _arrayEnd - 1;
         }
     }
 
@@ -1682,6 +1685,7 @@ private:
     // bsonArray
     const char* _arrayCurrent{nullptr};
     const char* _arrayEnd{nullptr};
+    size_t _fieldNameSize = 0;
 };
 
 /**
diff --git a/src/mongo/db/exec/sbe/vm/vm.cpp b/src/mongo/db/exec/sbe/vm/vm.cpp
index 3426a691ccf..0f2fa427ded 100644
--- a/src/mongo/db/exec/sbe/vm/vm.cpp
+++ b/src/mongo/db/exec/sbe/vm/vm.cpp
@@ -964,11 +964,11 @@ FastTuple<bool, value::TypeTags, value::Value> ByteCode::getField(value::TypeTag
         return {false, tag, val};
     } else if (objTag == value::TypeTags::bsonObject) {
         auto be = value::bitcastTo<const char*>(objValue);
-        auto end = be + ConstDataView(be).read<LittleEndian<uint32_t>>();
+        const auto end = be + ConstDataView(be).read<LittleEndian<uint32_t>>();
         // Skip document length.
         be += 4;
         while (be != end - 1) {
-            auto sv = bson::fieldNameView(be);
+            auto sv = bson::fieldNameAndLength(be);
 
             if (sv == fieldStr) {
                 auto [tag, val] = bson::convertFrom<true>(be, end, fieldStr.size());
@@ -1383,12 +1383,12 @@ FastTuple<bool, value::TypeTags, value::Value> ByteCode::setField() {
 
             if (objTag == value::TypeTags::bsonObject) {
                 auto be = value::bitcastTo<const char*>(objVal);
-                auto end = be + ConstDataView(be).read<LittleEndian<uint32_t>>();
+                const auto end = be + ConstDataView(be).read<LittleEndian<uint32_t>>();
 
                 // Skip document length.
                 be += 4;
-                while (*be != 0) {
-                    auto sv = bson::fieldNameView(be);
+                while (be != end - 1) {
+                    auto sv = bson::fieldNameAndLength(be);
 
                     if (sv != fieldName) {
                         auto [tag, val] = bson::convertFrom<false>(be, end, sv.size());
@@ -1424,12 +1424,12 @@ FastTuple<bool, value::TypeTags, value::Value> ByteCode::setField() {
 
         if (objTag == value::TypeTags::bsonObject) {
             auto be = value::bitcastTo<const char*>(objVal);
-            auto end = be + ConstDataView(be).read<LittleEndian<uint32_t>>();
+            const auto end = be + ConstDataView(be).read<LittleEndian<uint32_t>>();
 
             // Skip document length.
             be += 4;
-            while (*be != 0) {
-                auto sv = bson::fieldNameView(be);
+            while (be != end - 1) {
+                auto sv = bson::fieldNameAndLength(be);
 
                 if (sv != fieldName) {
                     auto [tag, val] = bson::convertFrom<false>(be, end, sv.size());
@@ -1908,11 +1908,11 @@ FastTuple<bool, value::TypeTags, value::Value> ByteCode::builtinDropFields(Arity
 
     if (tagInObj == value::TypeTags::bsonObject) {
         auto be = value::bitcastTo<const char*>(valInObj);
-        auto end = be + ConstDataView(be).read<LittleEndian<uint32_t>>();
+        const auto end = be + ConstDataView(be).read<LittleEndian<uint32_t>>();
         // Skip document length.
         be += 4;
-        while (*be != 0) {
-            auto sv = bson::fieldNameView(be);
+        while (be != end - 1) {
+            auto sv = bson::fieldNameAndLength(be);
 
             if (restrictFieldsSet.count(sv) == 0) {
                 auto [tag, val] = bson::convertFrom<false>(be, end, sv.size());
@@ -1984,11 +1984,11 @@ FastTuple<bool, value::TypeTags, value::Value> ByteCode::builtinKeepFields(Arity
 
     if (tagInObj == value::TypeTags::bsonObject) {
         auto be = value::bitcastTo<const char*>(valInObj);
-        auto end = be + ConstDataView(be).read<LittleEndian<uint32_t>>();
+        const auto end = be + ConstDataView(be).read<LittleEndian<uint32_t>>();
         // Skip document length.
         be += 4;
-        while (*be != 0) {
-            auto sv = bson::fieldNameView(be);
+        while (be != end - 1) {
+            auto sv = bson::fieldNameAndLength(be);
 
             if (keepFieldsSet.count(sv) == 1) {
                 auto [tag, val] = bson::convertFrom<false>(be, end, sv.size());
diff --git a/src/mongo/db/query/sbe_stage_builder_accumulator_test.cpp b/src/mongo/db/query/sbe_stage_builder_accumulator_test.cpp
index 8967f05841c..d98c96fe8a9 100644
--- a/src/mongo/db/query/sbe_stage_builder_accumulator_test.cpp
+++ b/src/mongo/db/query/sbe_stage_builder_accumulator_test.cpp
@@ -1892,7 +1892,7 @@ public:
         const char* bsonElt = valuesToAgg.objdata() + 4;
         const char* bsonEnd = bsonElt + valuesToAgg.objsize();
         while (*bsonElt != 0) {
-            auto fieldName = sbe::bson::fieldNameView(bsonElt);
+            auto fieldName = sbe::bson::fieldNameAndLength(bsonElt);
 
             // Convert the BSON value to an SBE value and put it inside the input slot.
             auto [tag, val] = sbe::bson::convertFrom<false>(bsonElt, bsonEnd, fieldName.size());