ESQL: Split large pages on load sometimes

benchmarks/README.md

@@ -152,11 +152,10 @@ exit
 Grab the async profiler from https://github.com/jvm-profiling-tools/async-profiler
 and run `prof async` like so:
 ```
-gradlew -p benchmarks/ run --args 'LongKeyedBucketOrdsBenchmark.multiBucket -prof "async:libPath=/home/nik9000/Downloads/async-profiler-3.0-29ee888-linux-x64/lib/libasyncProfiler.so;dir=/tmp/prof;output=flamegraph"'
+gradlew -p benchmarks/ run --args 'LongKeyedBucketOrdsBenchmark.multiBucket -prof "async:libPath=/home/nik9000/Downloads/async-profiler-4.0-linux-x64/lib/libasyncProfiler.so;dir=/tmp/prof;output=flamegraph"'
 ```
 
-Note: As of January 2025 the latest release of async profiler doesn't work
-      with our JDK but the nightly is fine.
+Note: As of July 2025 the 4.0 release of the async profiler works well.
 
 If you are on Mac, this'll warn you that you downloaded the shared library from
 the internet. You'll need to go to settings and allow it to run.

benchmarks/src/main/java/org/elasticsearch/benchmark/_nightly/esql/ValuesSourceReaderBenchmark.java

@@ -24,8 +24,10 @@
 import org.apache.lucene.util.BytesRef;
 import org.apache.lucene.util.NumericUtils;
 import org.elasticsearch.common.breaker.NoopCircuitBreaker;
+import org.elasticsearch.common.logging.LogConfigurator;
 import org.elasticsearch.common.lucene.Lucene;
 import org.elasticsearch.common.settings.Settings;
+import org.elasticsearch.common.unit.ByteSizeValue;
 import org.elasticsearch.common.util.BigArrays;
 import org.elasticsearch.compute.data.BlockFactory;
 import org.elasticsearch.compute.data.BytesRefBlock;
@@ -85,6 +87,10 @@
 @State(Scope.Thread)
 @Fork(1)
 public class ValuesSourceReaderBenchmark {
+    static {
+        LogConfigurator.configureESLogging();
+    }
+
     private static final String[] SUPPORTED_LAYOUTS = new String[] { "in_order", "shuffled", "shuffled_singles" };
     private static final String[] SUPPORTED_NAMES = new String[] {
         "long",
@@ -345,6 +351,7 @@ public FieldNamesFieldMapper.FieldNamesFieldType fieldNames() {
     public void benchmark() {
         ValuesSourceReaderOperator op = new ValuesSourceReaderOperator(
             blockFactory,
+            ByteSizeValue.ofMb(1).getBytes(),
             fields(name),
             List.of(new ValuesSourceReaderOperator.ShardContext(reader, () -> {
                 throw new UnsupportedOperationException("can't load _source here");

docs/changelog/131053.yaml

@@ -0,0 +1,5 @@
+pr: 131053
+summary: Split large pages on load sometimes
+area: ES|QL
+type: bug
+issues: []

server/src/main/java/org/elasticsearch/index/mapper/AbstractShapeGeometryFieldMapper.java

@@ -98,11 +98,11 @@ protected void writeExtent(BlockLoader.IntBuilder builder, Extent extent) {
             public BlockLoader.AllReader reader(LeafReaderContext context) throws IOException {
                 return new BlockLoader.AllReader() {
                     @Override
-                    public BlockLoader.Block read(BlockLoader.BlockFactory factory, BlockLoader.Docs docs) throws IOException {
+                    public BlockLoader.Block read(BlockLoader.BlockFactory factory, BlockLoader.Docs docs, int offset) throws IOException {
                         var binaryDocValues = context.reader().getBinaryDocValues(fieldName);
                         var reader = new GeometryDocValueReader();
-                        try (var builder = factory.ints(docs.count())) {
-                            for (int i = 0; i < docs.count(); i++) {
+                        try (var builder = factory.ints(docs.count() - offset)) {
+                            for (int i = offset; i < docs.count(); i++) {
                                 read(binaryDocValues, docs.get(i), reader, builder);
                             }
                             return builder.build();

server/src/main/java/org/elasticsearch/index/mapper/BlockDocValuesReader.java

@@ -124,10 +124,10 @@ private static class SingletonLongs extends BlockDocValuesReader {
         }
 
         @Override
-        public BlockLoader.Block read(BlockFactory factory, Docs docs) throws IOException {
-            try (BlockLoader.LongBuilder builder = factory.longsFromDocValues(docs.count())) {
+        public BlockLoader.Block read(BlockFactory factory, Docs docs, int offset) throws IOException {
+            try (BlockLoader.LongBuilder builder = factory.longsFromDocValues(docs.count() - offset)) {
                 int lastDoc = -1;
-                for (int i = 0; i < docs.count(); i++) {
+                for (int i = offset; i < docs.count(); i++) {
                     int doc = docs.get(i);
                     if (doc < lastDoc) {
                         throw new IllegalStateException("docs within same block must be in order");
@@ -173,9 +173,9 @@ private static class Longs extends BlockDocValuesReader {
         }
 
         @Override
-        public BlockLoader.Block read(BlockFactory factory, Docs docs) throws IOException {
-            try (BlockLoader.LongBuilder builder = factory.longsFromDocValues(docs.count())) {
-                for (int i = 0; i < docs.count(); i++) {
+        public BlockLoader.Block read(BlockFactory factory, Docs docs, int offset) throws IOException {
+            try (BlockLoader.LongBuilder builder = factory.longsFromDocValues(docs.count() - offset)) {
+                for (int i = offset; i < docs.count(); i++) {
                     int doc = docs.get(i);
                     if (doc < this.docID) {
                         throw new IllegalStateException("docs within same block must be in order");
@@ -259,10 +259,10 @@ private static class SingletonInts extends BlockDocValuesReader {
         }
 
         @Override
-        public BlockLoader.Block read(BlockFactory factory, Docs docs) throws IOException {
-            try (BlockLoader.IntBuilder builder = factory.intsFromDocValues(docs.count())) {
+        public BlockLoader.Block read(BlockFactory factory, Docs docs, int offset) throws IOException {
+            try (BlockLoader.IntBuilder builder = factory.intsFromDocValues(docs.count() - offset)) {
                 int lastDoc = -1;
-                for (int i = 0; i < docs.count(); i++) {
+                for (int i = offset; i < docs.count(); i++) {
                     int doc = docs.get(i);
                     if (doc < lastDoc) {
                         throw new IllegalStateException("docs within same block must be in order");
@@ -308,9 +308,9 @@ private static class Ints extends BlockDocValuesReader {
         }
 
         @Override
-        public BlockLoader.Block read(BlockFactory factory, Docs docs) throws IOException {
-            try (BlockLoader.IntBuilder builder = factory.intsFromDocValues(docs.count())) {
-                for (int i = 0; i < docs.count(); i++) {
+        public BlockLoader.Block read(BlockFactory factory, Docs docs, int offset) throws IOException {
+            try (BlockLoader.IntBuilder builder = factory.intsFromDocValues(docs.count() - offset)) {
+                for (int i = offset; i < docs.count(); i++) {
                     int doc = docs.get(i);
                     if (doc < this.docID) {
                         throw new IllegalStateException("docs within same block must be in order");
@@ -408,10 +408,10 @@ private static class SingletonDoubles extends BlockDocValuesReader {
         }
 
         @Override
-        public BlockLoader.Block read(BlockFactory factory, Docs docs) throws IOException {
-            try (BlockLoader.DoubleBuilder builder = factory.doublesFromDocValues(docs.count())) {
+        public BlockLoader.Block read(BlockFactory factory, Docs docs, int offset) throws IOException {
+            try (BlockLoader.DoubleBuilder builder = factory.doublesFromDocValues(docs.count() - offset)) {
                 int lastDoc = -1;
-                for (int i = 0; i < docs.count(); i++) {
+                for (int i = offset; i < docs.count(); i++) {
                     int doc = docs.get(i);
                     if (doc < lastDoc) {
                         throw new IllegalStateException("docs within same block must be in order");
@@ -461,9 +461,9 @@ private static class Doubles extends BlockDocValuesReader {
         }
 
         @Override
-        public BlockLoader.Block read(BlockFactory factory, Docs docs) throws IOException {
-            try (BlockLoader.DoubleBuilder builder = factory.doublesFromDocValues(docs.count())) {
-                for (int i = 0; i < docs.count(); i++) {
+        public BlockLoader.Block read(BlockFactory factory, Docs docs, int offset) throws IOException {
+            try (BlockLoader.DoubleBuilder builder = factory.doublesFromDocValues(docs.count() - offset)) {
+                for (int i = offset; i < docs.count(); i++) {
                     int doc = docs.get(i);
                     if (doc < this.docID) {
                         throw new IllegalStateException("docs within same block must be in order");
@@ -544,10 +544,10 @@ private static class DenseVectorValuesBlockReader extends BlockDocValuesReader {
         }
 
         @Override
-        public BlockLoader.Block read(BlockFactory factory, Docs docs) throws IOException {
+        public BlockLoader.Block read(BlockFactory factory, Docs docs, int offset) throws IOException {
             // Doubles from doc values ensures that the values are in order
             try (BlockLoader.FloatBuilder builder = factory.denseVectors(docs.count(), dimensions)) {
-                for (int i = 0; i < docs.count(); i++) {
+                for (int i = offset; i < docs.count(); i++) {
                     int doc = docs.get(i);
                     if (doc < iterator.docID()) {
                         throw new IllegalStateException("docs within same block must be in order");
@@ -645,19 +645,19 @@ private BlockLoader.Block readSingleDoc(BlockFactory factory, int docId) throws
             if (ordinals.advanceExact(docId)) {
                 BytesRef v = ordinals.lookupOrd(ordinals.ordValue());
                 // the returned BytesRef can be reused
-                return factory.constantBytes(BytesRef.deepCopyOf(v));
+                return factory.constantBytes(BytesRef.deepCopyOf(v), 1);
             } else {
-                return factory.constantNulls();
+                return factory.constantNulls(1);
             }
         }
 
         @Override
-        public BlockLoader.Block read(BlockFactory factory, Docs docs) throws IOException {
-            if (docs.count() == 1) {
-                return readSingleDoc(factory, docs.get(0));
+        public BlockLoader.Block read(BlockFactory factory, Docs docs, int offset) throws IOException {
+            if (docs.count() - offset == 1) {
+                return readSingleDoc(factory, docs.get(offset));
             }
-            try (var builder = factory.singletonOrdinalsBuilder(ordinals, docs.count())) {
-                for (int i = 0; i < docs.count(); i++) {
+            try (var builder = factory.singletonOrdinalsBuilder(ordinals, docs.count() - offset)) {
+                for (int i = offset; i < docs.count(); i++) {
                     int doc = docs.get(i);
                     if (doc < ordinals.docID()) {
                         throw new IllegalStateException("docs within same block must be in order");
@@ -700,12 +700,12 @@ private static class Ordinals extends BlockDocValuesReader {
         }
 
         @Override
-        public BlockLoader.Block read(BlockFactory factory, Docs docs) throws IOException {
-            if (docs.count() == 1) {
-                return readSingleDoc(factory, docs.get(0));
+        public BlockLoader.Block read(BlockFactory factory, Docs docs, int offset) throws IOException {
+            if (docs.count() - offset == 1) {
+                return readSingleDoc(factory, docs.get(offset));
             }
-            try (var builder = factory.sortedSetOrdinalsBuilder(ordinals, docs.count())) {
-                for (int i = 0; i < docs.count(); i++) {
+            try (var builder = factory.sortedSetOrdinalsBuilder(ordinals, docs.count() - offset)) {
+                for (int i = offset; i < docs.count(); i++) {
                     int doc = docs.get(i);
                     if (doc < ordinals.docID()) {
                         throw new IllegalStateException("docs within same block must be in order");
@@ -736,12 +736,12 @@ public void read(int docId, BlockLoader.StoredFields storedFields, Builder build
 
         private BlockLoader.Block readSingleDoc(BlockFactory factory, int docId) throws IOException {
             if (ordinals.advanceExact(docId) == false) {
-                return factory.constantNulls();
+                return factory.constantNulls(1);
             }
             int count = ordinals.docValueCount();
             if (count == 1) {
                 BytesRef v = ordinals.lookupOrd(ordinals.nextOrd());
-                return factory.constantBytes(BytesRef.deepCopyOf(v));
+                return factory.constantBytes(BytesRef.deepCopyOf(v), 1);
             }
             try (var builder = factory.bytesRefsFromDocValues(count)) {
                 builder.beginPositionEntry();
@@ -816,9 +816,9 @@ private static class BytesRefsFromBinary extends BlockDocValuesReader {
         }
 
         @Override
-        public BlockLoader.Block read(BlockFactory factory, Docs docs) throws IOException {
-            try (BlockLoader.BytesRefBuilder builder = factory.bytesRefs(docs.count())) {
-                for (int i = 0; i < docs.count(); i++) {
+        public BlockLoader.Block read(BlockFactory factory, Docs docs, int offset) throws IOException {
+            try (BlockLoader.BytesRefBuilder builder = factory.bytesRefs(docs.count() - offset)) {
+                for (int i = offset; i < docs.count(); i++) {
                     int doc = docs.get(i);
                     if (doc < docID) {
                         throw new IllegalStateException("docs within same block must be in order");
@@ -915,9 +915,9 @@ private static class DenseVectorFromBinary extends BlockDocValuesReader {
         }
 
         @Override
-        public BlockLoader.Block read(BlockFactory factory, Docs docs) throws IOException {
+        public BlockLoader.Block read(BlockFactory factory, Docs docs, int offset) throws IOException {
             try (BlockLoader.FloatBuilder builder = factory.denseVectors(docs.count(), dimensions)) {
-                for (int i = 0; i < docs.count(); i++) {
+                for (int i = offset; i < docs.count(); i++) {
                     int doc = docs.get(i);
                     if (doc < docID) {
                         throw new IllegalStateException("docs within same block must be in order");
@@ -999,10 +999,10 @@ private static class SingletonBooleans extends BlockDocValuesReader {
         }
 
         @Override
-        public BlockLoader.Block read(BlockFactory factory, Docs docs) throws IOException {
-            try (BlockLoader.BooleanBuilder builder = factory.booleansFromDocValues(docs.count())) {
+        public BlockLoader.Block read(BlockFactory factory, Docs docs, int offset) throws IOException {
+            try (BlockLoader.BooleanBuilder builder = factory.booleansFromDocValues(docs.count() - offset)) {
                 int lastDoc = -1;
-                for (int i = 0; i < docs.count(); i++) {
+                for (int i = offset; i < docs.count(); i++) {
                     int doc = docs.get(i);
                     if (doc < lastDoc) {
                         throw new IllegalStateException("docs within same block must be in order");
@@ -1048,9 +1048,9 @@ private static class Booleans extends BlockDocValuesReader {
         }
 
         @Override
-        public BlockLoader.Block read(BlockFactory factory, Docs docs) throws IOException {
-            try (BlockLoader.BooleanBuilder builder = factory.booleansFromDocValues(docs.count())) {
-                for (int i = 0; i < docs.count(); i++) {
+        public BlockLoader.Block read(BlockFactory factory, Docs docs, int offset) throws IOException {
+            try (BlockLoader.BooleanBuilder builder = factory.booleansFromDocValues(docs.count() - offset)) {
+                for (int i = offset; i < docs.count(); i++) {
                     int doc = docs.get(i);
                     if (doc < this.docID) {
                         throw new IllegalStateException("docs within same block must be in order");

server/src/main/java/org/elasticsearch/index/mapper/BlockLoader.java

@@ -43,7 +43,7 @@ interface ColumnAtATimeReader extends Reader {
         /**
          * Reads the values of all documents in {@code docs}.
          */
-        BlockLoader.Block read(BlockFactory factory, Docs docs) throws IOException;
+        BlockLoader.Block read(BlockFactory factory, Docs docs, int offset) throws IOException;
     }
 
     interface RowStrideReader extends Reader {
@@ -149,8 +149,8 @@ public String toString() {
      */
     class ConstantNullsReader implements AllReader {
         @Override
-        public Block read(BlockFactory factory, Docs docs) throws IOException {
-            return factory.constantNulls();
+        public Block read(BlockFactory factory, Docs docs, int offset) throws IOException {
+            return factory.constantNulls(docs.count() - offset);
         }
 
         @Override
@@ -183,8 +183,8 @@ public Builder builder(BlockFactory factory, int expectedCount) {
             public ColumnAtATimeReader columnAtATimeReader(LeafReaderContext context) {
                 return new ColumnAtATimeReader() {
                     @Override
-                    public Block read(BlockFactory factory, Docs docs) {
-                        return factory.constantBytes(value);
+                    public Block read(BlockFactory factory, Docs docs, int offset) {
+                        return factory.constantBytes(value, docs.count() - offset);
                     }
 
                     @Override
@@ -261,8 +261,8 @@ public ColumnAtATimeReader columnAtATimeReader(LeafReaderContext context) throws
             }
             return new ColumnAtATimeReader() {
                 @Override
-                public Block read(BlockFactory factory, Docs docs) throws IOException {
-                    return reader.read(factory, docs);
+                public Block read(BlockFactory factory, Docs docs, int offset) throws IOException {
+                    return reader.read(factory, docs, offset);
                 }
 
                 @Override
@@ -408,13 +408,13 @@ interface BlockFactory {
         /**
          * Build a block that contains only {@code null}.
          */
-        Block constantNulls();
+        Block constantNulls(int count);
 
         /**
          * Build a block that contains {@code value} repeated
          * {@code size} times.
          */
-        Block constantBytes(BytesRef value);
+        Block constantBytes(BytesRef value, int count);
 
         /**
          * Build a reader for reading {@link SortedDocValues}

server/src/main/java/org/elasticsearch/index/mapper/BooleanScriptBlockDocValuesReader.java

@@ -49,10 +49,10 @@ public int docId() {
     }
 
     @Override
-    public BlockLoader.Block read(BlockLoader.BlockFactory factory, BlockLoader.Docs docs) throws IOException {
+    public BlockLoader.Block read(BlockLoader.BlockFactory factory, BlockLoader.Docs docs, int offset) throws IOException {
         // Note that we don't emit falses before trues so we conform to the doc values contract and can use booleansFromDocValues
-        try (BlockLoader.BooleanBuilder builder = factory.booleans(docs.count())) {
-            for (int i = 0; i < docs.count(); i++) {
+        try (BlockLoader.BooleanBuilder builder = factory.booleans(docs.count() - offset)) {
+            for (int i = offset; i < docs.count(); i++) {
                 read(docs.get(i), builder);
             }
             return builder.build();

server/src/main/java/org/elasticsearch/index/mapper/DateScriptBlockDocValuesReader.java

@@ -49,10 +49,10 @@ public int docId() {
     }
 
     @Override
-    public BlockLoader.Block read(BlockLoader.BlockFactory factory, BlockLoader.Docs docs) throws IOException {
+    public BlockLoader.Block read(BlockLoader.BlockFactory factory, BlockLoader.Docs docs, int offset) throws IOException {
         // Note that we don't sort the values sort, so we can't use factory.longsFromDocValues
-        try (BlockLoader.LongBuilder builder = factory.longs(docs.count())) {
-            for (int i = 0; i < docs.count(); i++) {
+        try (BlockLoader.LongBuilder builder = factory.longs(docs.count() - offset)) {
+            for (int i = offset; i < docs.count(); i++) {
                 read(docs.get(i), builder);
             }
             return builder.build();