Merge branch 'master' into godriver2388v2

Author: qingyang-hu

.evergreen/config.yml

@@ -388,6 +388,14 @@ functions:
           TASKFILE_TARGET: test-short
         args: [*task-runner, run-docker]
 
+  run-goleak-test:
+    - command: subprocess.exec
+      type: test
+      params:
+        binary: "bash"
+        include_expansions_in_env: ["MONGODB_URI"]
+        args: [*task-runner, test-goleak]
+
   run-ocsp-server:
     - command: subprocess.exec
       params:
@@ -1100,6 +1108,26 @@ tasks:
       - func: bootstrap-mongo-orchestration
       - func: run-docker-test
 
+  - name: test-goroutine-leaks-replicaset
+    tags: ["goleak"]
+    commands:
+      - func: bootstrap-mongo-orchestration
+        vars:
+          TOPOLOGY: "replica_set"
+          AUTH: "noauth"
+          SSL: "nossl"
+      - func: run-goleak-test
+
+  - name: test-goroutine-leaks-sharded
+    tags: ["goleak"]
+    commands:
+      - func: bootstrap-mongo-orchestration
+        vars:
+          TOPOLOGY: "sharded_cluster"
+          AUTH: "noauth"
+          SSL: "nossl"
+      - func: run-goleak-test
+
   - name: test-load-balancer-noauth-nossl
     tags: ["load-balancer"]
     commands:
@@ -2056,6 +2084,16 @@ buildvariants:
     tasks:
       - name: "test-docker-runner"
 
+  - name: goroutine-leaks-test
+    tags: ["pullrequest"]
+    display_name: "Goroutine Leaks Test"
+    run_on:
+      - ubuntu2204-large
+    expansions:
+      GO_DIST: "/opt/golang/go1.22"
+    tasks:
+      - name: ".goleak"
+
   - matrix_name: "tests-rhel-36-with-zlib-support"
     tags: ["pullrequest"]
     matrix_spec: { version: ["3.6"], os-ssl-32: ["rhel87-64"] }

Dockerfile

@@ -1,7 +1,7 @@
 # Dockerfile for Go Driver local development.
 
 # Build libmongocrypt in a separate build stage.
-FROM ubuntu:20.04 as libmongocrypt
+FROM artifactory.corp.mongodb.com/dockerhub/ubuntu:20.04 as libmongocrypt
 
 RUN apt-get -qq update && \
   apt-get -qqy install --no-install-recommends \
@@ -17,7 +17,7 @@ RUN cd /root && bash ./install-libmongocrypt.sh
 
 
 # Copy in the files from the libmongocrypt build stage.
-FROM ubuntu:20.04
+FROM artifactory.corp.mongodb.com/dockerhub/ubuntu:20.04
 
 # Install common deps.
 RUN export DEBIAN_FRONTEND=noninteractive && \

Taskfile.yml

@@ -105,6 +105,8 @@ tasks:
 
   test-gcpkms: bash etc/run-gcpkms-test.sh
 
+  test-goleak: bash etc/run-goleak-test.sh
+
   ### Local FaaS tasks. ###
   build-faas-awslambda:
     requires: 

bson/decoder.go

@@ -87,7 +87,7 @@ func (d *Decoder) SetRegistry(r *Registry) {
 	d.dc.Registry = r
 }
 
-// DefaultDocumentM causes the Decoder to always unmarshal documents into the primitive.M type. This
+// DefaultDocumentM causes the Decoder to always unmarshal documents into the bson.M type. This
 // behavior is restricted to data typed as "interface{}" or "map[string]interface{}".
 func (d *Decoder) DefaultDocumentM() {
 	d.dc.defaultDocumentType = reflect.TypeOf(M{})
@@ -101,7 +101,7 @@ func (d *Decoder) AllowTruncatingDoubles() {
 }
 
 // BinaryAsSlice causes the Decoder to unmarshal BSON binary field values that are the "Generic" or
-// "Old" BSON binary subtype as a Go byte slice instead of a primitive.Binary.
+// "Old" BSON binary subtype as a Go byte slice instead of a bson.Binary.
 func (d *Decoder) BinaryAsSlice() {
 	d.dc.binaryAsSlice = true
 }

bson/decoder_test.go

@@ -647,7 +647,8 @@ func TestDecoderConfiguration(t *testing.T) {
 
 		var got objectIDTest
 		err := dec.Decode(&got)
-		assert.EqualError(t, err, "error decoding key id: decoding an object ID to a non-hexadecimal string representation is not supported")
+		const want = "error decoding key id: decoding an object ID into a string is not supported by default (set Decoder.ObjectIDAsHexString to enable decoding as a hexadecimal string)"
+		assert.EqualError(t, err, want)
 	})
 	t.Run("DefaultDocumentM top-level", func(t *testing.T) {
 		t.Parallel()

bson/doc.go

@@ -47,20 +47,20 @@
 //  5. BSON boolean unmarshals to a bool.
 //  6. BSON embedded document unmarshals to the parent type (i.e. D for a D, M for an M).
 //  7. BSON array unmarshals to a bson.A.
-//  8. BSON ObjectId unmarshals to a primitive.ObjectID.
-//  9. BSON datetime unmarshals to a primitive.DateTime.
-//  10. BSON binary unmarshals to a primitive.Binary.
-//  11. BSON regular expression unmarshals to a primitive.Regex.
-//  12. BSON JavaScript unmarshals to a primitive.JavaScript.
-//  13. BSON code with scope unmarshals to a primitive.CodeWithScope.
-//  14. BSON timestamp unmarshals to an primitive.Timestamp.
-//  15. BSON 128-bit decimal unmarshals to an primitive.Decimal128.
-//  16. BSON min key unmarshals to an primitive.MinKey.
-//  17. BSON max key unmarshals to an primitive.MaxKey.
-//  18. BSON undefined unmarshals to a primitive.Undefined.
+//  8. BSON ObjectId unmarshals to a bson.ObjectID.
+//  9. BSON datetime unmarshals to a bson.DateTime.
+//  10. BSON binary unmarshals to a bson.Binary.
+//  11. BSON regular expression unmarshals to a bson.Regex.
+//  12. BSON JavaScript unmarshals to a bson.JavaScript.
+//  13. BSON code with scope unmarshals to a bson.CodeWithScope.
+//  14. BSON timestamp unmarshals to an bson.Timestamp.
+//  15. BSON 128-bit decimal unmarshals to an bson.Decimal128.
+//  16. BSON min key unmarshals to an bson.MinKey.
+//  17. BSON max key unmarshals to an bson.MaxKey.
+//  18. BSON undefined unmarshals to a bson.Undefined.
 //  19. BSON null unmarshals to nil.
-//  20. BSON DBPointer unmarshals to a primitive.DBPointer.
-//  21. BSON symbol unmarshals to a primitive.Symbol.
+//  20. BSON DBPointer unmarshals to a bson.DBPointer.
+//  21. BSON symbol unmarshals to a bson.Symbol.
 //
 // The above mappings also apply when marshaling a D or M to BSON. Some other useful marshaling mappings are:
 //

bson/primitive.go

@@ -60,7 +60,7 @@ func (d DateTime) MarshalJSON() ([]byte, error) {
 	return d.Time().UTC().MarshalJSON()
 }
 
-// UnmarshalJSON creates a primitive.DateTime from a JSON string.
+// UnmarshalJSON creates a bson.DateTime from a JSON string.
 func (d *DateTime) UnmarshalJSON(data []byte) error {
 	// Ignore "null" so that we can distinguish between a "null" value and
 	// valid value that is the zero time (as reported by time.Time.IsZero).

bson/string_codec.go

@@ -7,6 +7,7 @@
 package bson
 
 import (
+	"errors"
 	"fmt"
 	"reflect"
 )
@@ -50,14 +51,16 @@ func (sc *stringCodec) decodeType(dc DecodeContext, vr ValueReader, t reflect.Ty
 			return emptyValue, err
 		}
 	case TypeObjectID:
-		oid, err := vr.ReadObjectID()
-		if err != nil {
-			return emptyValue, err
-		}
 		if dc.objectIDAsHexString {
+			oid, err := vr.ReadObjectID()
+			if err != nil {
+				return emptyValue, err
+			}
 			str = oid.Hex()
 		} else {
-			return emptyValue, fmt.Errorf("decoding an object ID to a non-hexadecimal string representation is not supported")
+			const msg = "decoding an object ID into a string is not supported by default " +
+				"(set Decoder.ObjectIDAsHexString to enable decoding as a hexadecimal string)"
+			return emptyValue, errors.New(msg)
 		}
 	case TypeSymbol:
 		str, err = vr.ReadSymbol()

docs/migration-2.0.md

@@ -867,6 +867,34 @@ The `NewRegistryBuilder` function has been removed along with the `bsoncodec.Reg
 
 ### Decoder
 
+The BSON decoding logic has changed to decode into a `bson.D` by default.
+
+The example shows the behavior change.
+
+```go
+// v1
+
+b1 := bson.M{"a": 1, "b": bson.M{"c": 2}}
+b2, _ := bson.Marshal(b1)
+b3 := bson.M{}
+bson.Unmarshal(b2, &b3)
+fmt.Printf("b3.b type: %T\n", b3["b"])
+// Output: b3.b type: primitive.M
+```
+
+```go
+// v2
+
+b1 := bson.M{"a": 1, "b": bson.M{"c": 2}}
+b2, _ := bson.Marshal(b1)
+b3 := bson.M{}
+bson.Unmarshal(b2, &b3)
+fmt.Printf("b3.b type: %T\n", b3["b"])
+// Output: b3.b type: bson.D
+```
+
+Use `Decoder.DefaultDocumentM()` or set the `DefaultDocumentM` field of `options.BSONOptions` to always decode documents into the `bson.M` type.
+
 #### NewDecoder
 
 The signature of `NewDecoder` has been updated without an error being returned.
@@ -1043,11 +1071,11 @@ The signature of `Reset` has been updated without an error being returned.
 
 #### DefaultDocumentD / DefaultDocumentM
 
-`Decoder.DefaultDocumentD` has been removed since a document, including a top-level value (e.g. you pass in an empty interface value to Decode), is always decoded into a `bson.D` by default. Therefore, use `Decoder.DefaultDocumentM` to always decode a document into a `bson.M` to avoid unexpected decode results.
+`Decoder.DefaultDocumentD()` has been removed since a document, including a top-level value (e.g. you pass in an empty interface value to Decode), is always decoded into a `bson.D` by default. Therefore, use `Decoder.DefaultDocumentM()` to always decode a document into a `bson.M` to avoid unexpected decode results.
 
 #### ObjectIDAsHexString
 
-`Decoder.ObjectIDAsHexString` method enables decoding a BSON ObjectId as a hexadecimal string. Otherwise, the decoder returns an error by default instead of decoding as the UTF-8 representation of the raw ObjectId bytes, which results in a garbled and unusable string.
+`Decoder.ObjectIDAsHexString()` method enables decoding a BSON ObjectId as a hexadecimal string. Otherwise, the decoder returns an error by default instead of decoding as the UTF-8 representation of the raw ObjectId bytes, which results in a garbled and unusable string.
 
 ### Encoder
 
@@ -1183,8 +1211,20 @@ A new `RawArray` type has been added to the `bson` package as a primitive type t
 
 ### ValueMarshaler
 
-The `MarshalBSONValue` method of the `ValueMarshaler` interface is only required to return a byte type value representing the BSON type to avoid importing the `bsontype` package.
+The `MarshalBSONValue` method of the [ValueMarshaler](https://pkg.go.dev/go.mongodb.org/mongo-driver/v2/bson#ValueMarshaler) interface now returns a `byte` value representing the [BSON type](https://pkg.go.dev/go.mongodb.org/mongo-driver/v2/bson#Type). That allows external packages to implement the `ValueMarshaler` interface without having to import the `bson` package. Convert a returned `byte` value to [bson.Type](https://pkg.go.dev/go.mongodb.org/mongo-driver/v2/bson#Type) to compare with the BSON type constants. For example:
+
+```go
+btype, _, _ := m.MarshalBSONValue()
+fmt.Println("type of data: %s: ", bson.Type(btype))
+fmt.Println("type of data is an array: %v", bson.Type(btype) == bson.TypeArray)
+```
 
 ### ValueUnmarshaler
 
-The `UnmarshalBSONValue` method of the `ValueUnmarshaler` interface is only required to take a byte type argument representing the BSON type to avoid importing the Go driver package.
+The `UnmarshalBSONValue` method of the [ValueUnmarshaler](https://pkg.go.dev/go.mongodb.org/mongo-driver/v2/bson#ValueUnmarshaler) interface now accepts a `byte` value representing the [BSON type](https://pkg.go.dev/go.mongodb.org/mongo-driver/v2/bson#Type) for the first argument. That allows packages to implement `ValueUnmarshaler` without having to import the `bson` package. For example:
+
+```go
+if err := m.UnmarshalBSONValue(bson.TypeEmbeddedDocument, bytes); err != nil {
+    log.Fatalf("failed to decode embedded document: %v", err)
+}
+```

etc/run-goleak-test.sh

@@ -0,0 +1,10 @@
+#!/usr/bin/env bash
+# run-goleak-test
+# Run goroutine leak tests.
+set -eu
+set +x
+
+echo "Running internal/test/goleak"
+pushd internal/test/goleak
+go test -v ./... >> ../../../test.suite
+popd