This is an automated cherry-pick of #47637

Signed-off-by: ti-chi-bot <[email protected]>

pkg/executor/BUILD.bazel

@@ -256,6 +256,7 @@ go_library(
         "@com_github_prometheus_client_golang//api/prometheus/v1:prometheus",
         "@com_github_prometheus_client_golang//prometheus",
         "@com_github_prometheus_common//model",
+        "@com_github_tiancaiamao_gp//:gp",
         "@com_github_tikv_client_go_v2//error",
         "@com_github_tikv_client_go_v2//kv",
         "@com_github_tikv_client_go_v2//oracle",

pkg/executor/analyze.go

@@ -47,6 +47,7 @@ import (
 	"github.com/pingcap/tidb/pkg/util/logutil"
 	"github.com/pingcap/tidb/pkg/util/sqlexec"
 	"github.com/pingcap/tipb/go-tipb"
+	"github.com/tiancaiamao/gp"
 	"go.uber.org/zap"
 	"golang.org/x/sync/errgroup"
 )
@@ -57,9 +58,10 @@ var _ exec.Executor = &AnalyzeExec{}
 type AnalyzeExec struct {
 	exec.BaseExecutor
 	tasks      []*analyzeTask
-	wg         util.WaitGroupWrapper
+	wg         *util.WaitGroupPool
 	opts       map[ast.AnalyzeOptionType]uint64
 	OptionsMap map[int64]core.V2AnalyzeOptions
+	gp         *gp.Pool
 }
 
 var (
@@ -419,7 +421,7 @@ func (e *AnalyzeExec) handleResultsErrorWithConcurrency(ctx context.Context, sta
 	globalStatsMap globalStatsMap, resultsCh <-chan *statistics.AnalyzeResults) error {
 	partitionStatsConcurrency := len(subSctxs)
 
-	var wg util.WaitGroupWrapper
+	wg := util.NewWaitGroupPool(e.gp)
 	saveResultsCh := make(chan *statistics.AnalyzeResults, partitionStatsConcurrency)
 	errCh := make(chan error, partitionStatsConcurrency)
 	for i := 0; i < partitionStatsConcurrency; i++ {
@@ -495,7 +497,7 @@ func (e *AnalyzeExec) analyzeWorker(taskCh <-chan *analyzeTask, resultsCh chan<-
 		StartAnalyzeJob(e.Ctx(), task.job)
 		switch task.taskType {
 		case colTask:
-			resultsCh <- analyzeColumnsPushDownEntry(task.colExec)
+			resultsCh <- analyzeColumnsPushDownEntry(e.gp, task.colExec)
 		case idxTask:
 			resultsCh <- analyzeIndexPushdown(task.idxExec)
 		}

pkg/executor/analyze_col.go

@@ -40,6 +40,7 @@ import (
 	"github.com/pingcap/tidb/pkg/util/memory"
 	"github.com/pingcap/tidb/pkg/util/ranger"
 	"github.com/pingcap/tipb/go-tipb"
+	"github.com/tiancaiamao/gp"
 )
 
 // AnalyzeColumnsExec represents Analyze columns push down executor.
@@ -64,9 +65,9 @@ type AnalyzeColumnsExec struct {
 	memTracker *memory.Tracker
 }
 
-func analyzeColumnsPushDownEntry(e *AnalyzeColumnsExec) *statistics.AnalyzeResults {
+func analyzeColumnsPushDownEntry(gp *gp.Pool, e *AnalyzeColumnsExec) *statistics.AnalyzeResults {
 	if e.AnalyzeInfo.StatsVersion >= statistics.Version2 {
-		return e.toV2().analyzeColumnsPushDownWithRetryV2()
+		return e.toV2().analyzeColumnsPushDownWithRetryV2(gp)
 	}
 	return e.toV1().analyzeColumnsPushDownV1()
 }

pkg/executor/analyze_col_v2.go

@@ -46,6 +46,7 @@ import (
 	"github.com/pingcap/tidb/pkg/util/ranger"
 	"github.com/pingcap/tidb/pkg/util/timeutil"
 	"github.com/pingcap/tipb/go-tipb"
+	"github.com/tiancaiamao/gp"
 	"go.uber.org/zap"
 	"golang.org/x/sync/errgroup"
 )
@@ -55,8 +56,8 @@ type AnalyzeColumnsExecV2 struct {
 	*AnalyzeColumnsExec
 }
 
-func (e *AnalyzeColumnsExecV2) analyzeColumnsPushDownWithRetryV2() *statistics.AnalyzeResults {
-	analyzeResult := e.analyzeColumnsPushDownV2()
+func (e *AnalyzeColumnsExecV2) analyzeColumnsPushDownWithRetryV2(gp *gp.Pool) *statistics.AnalyzeResults {
+	analyzeResult := e.analyzeColumnsPushDownV2(gp)
 	if e.notRetryable(analyzeResult) {
 		return analyzeResult
 	}
@@ -86,7 +87,7 @@ func (e *AnalyzeColumnsExecV2) analyzeColumnsPushDownWithRetryV2() *statistics.A
 	prepareV2AnalyzeJobInfo(e.AnalyzeColumnsExec, true)
 	AddNewAnalyzeJob(e.ctx, e.job)
 	StartAnalyzeJob(e.ctx, e.job)
-	return e.analyzeColumnsPushDownV2()
+	return e.analyzeColumnsPushDownV2(gp)
 }
 
 // Do **not** retry if succeed / not oom error / not auto-analyze / samplerate not set.
@@ -96,7 +97,7 @@ func (e *AnalyzeColumnsExecV2) notRetryable(analyzeResult *statistics.AnalyzeRes
 		e.analyzePB.ColReq == nil || *e.analyzePB.ColReq.SampleRate <= 0
 }
 
-func (e *AnalyzeColumnsExecV2) analyzeColumnsPushDownV2() *statistics.AnalyzeResults {
+func (e *AnalyzeColumnsExecV2) analyzeColumnsPushDownV2(gp *gp.Pool) *statistics.AnalyzeResults {
 	var ranges []*ranger.Range
 	if hc := e.handleCols; hc != nil {
 		if hc.IsInt() {
@@ -141,13 +142,13 @@ func (e *AnalyzeColumnsExecV2) analyzeColumnsPushDownV2() *statistics.AnalyzeRes
 	// subIndexWorkerWg is better to be initialized in handleNDVForSpecialIndexes, however if we do so, golang would
 	// report unexpected/unreasonable data race error on subIndexWorkerWg when running TestAnalyzeVirtualCol test
 	// case with `-race` flag now.
-	var wg util.WaitGroupWrapper
+	wg := util.NewWaitGroupPool(gp)
 	wg.Run(func() {
 		e.handleNDVForSpecialIndexes(specialIndexes, idxNDVPushDownCh, statsConcurrncy)
 	})
 	defer wg.Wait()
 
-	count, hists, topNs, fmSketches, extStats, err := e.buildSamplingStats(ranges, collExtStats, specialIndexesOffsets, idxNDVPushDownCh, samplingStatsConcurrency)
+	count, hists, topNs, fmSketches, extStats, err := e.buildSamplingStats(gp, ranges, collExtStats, specialIndexesOffsets, idxNDVPushDownCh, samplingStatsConcurrency)
 	if err != nil {
 		e.memTracker.Release(e.memTracker.BytesConsumed())
 		return &statistics.AnalyzeResults{Err: err, Job: e.job}
@@ -245,6 +246,7 @@ func printAnalyzeMergeCollectorLog(oldRootCount, newRootCount, subCount, tableID
 }
 
 func (e *AnalyzeColumnsExecV2) buildSamplingStats(
+	gp *gp.Pool,
 	ranges []*ranger.Range,
 	needExtStats bool,
 	indexesWithVirtualColOffsets []int,
@@ -292,7 +294,10 @@ func (e *AnalyzeColumnsExecV2) buildSamplingStats(
 	e.samplingMergeWg = &util.WaitGroupWrapper{}
 	e.samplingMergeWg.Add(samplingStatsConcurrency)
 	for i := 0; i < samplingStatsConcurrency; i++ {
-		go e.subMergeWorker(mergeResultCh, mergeTaskCh, l, i)
+		id := i
+		gp.Go(func() {
+			e.subMergeWorker(mergeResultCh, mergeTaskCh, l, id)
+		})
 	}
 	// Merge the result from collectors.
 	mergeWorkerPanicCnt := 0
@@ -388,7 +393,7 @@ func (e *AnalyzeColumnsExecV2) buildSamplingStats(
 	if totalLen < samplingStatsConcurrency {
 		samplingStatsConcurrency = totalLen
 	}
-	e.samplingBuilderWg = newNotifyErrorWaitGroupWrapper(buildResultChan)
+	e.samplingBuilderWg = newNotifyErrorWaitGroupWrapper(gp, buildResultChan)
 	sampleCollectors := make([]*statistics.SampleCollector, len(e.colsInfo))
 	exitCh := make(chan struct{})
 	e.samplingBuilderWg.Add(samplingStatsConcurrency)

pkg/executor/analyze_utils.go

@@ -25,7 +25,12 @@ import (
 	"github.com/pingcap/tidb/pkg/sessionctx"
 	"github.com/pingcap/tidb/pkg/sessionctx/variable"
 	"github.com/pingcap/tidb/pkg/statistics"
+<<<<<<< HEAD
 	"github.com/pingcap/tidb/pkg/util/memory"
+=======
+	"github.com/pingcap/tidb/pkg/util"
+	"github.com/tiancaiamao/gp"
+>>>>>>> 5a22d566a88 (executor: reuse goroutine in the analyze (#47637))
 	"go.uber.org/atomic"
 )
 
@@ -107,16 +112,17 @@ func (w *analyzeResultsNotifyWaitGroupWrapper) Run(exec func()) {
 // notifyErrorWaitGroupWrapper is a wrapper for sync.WaitGroup
 // Please add all goroutine count when to `Add` to avoid exiting in advance.
 type notifyErrorWaitGroupWrapper struct {
-	sync.WaitGroup
+	*util.WaitGroupPool
 	notify chan error
 	cnt    atomic.Uint64
 }
 
 // newNotifyErrorWaitGroupWrapper is to create notifyErrorWaitGroupWrapper
-func newNotifyErrorWaitGroupWrapper(notify chan error) *notifyErrorWaitGroupWrapper {
+func newNotifyErrorWaitGroupWrapper(gp *gp.Pool, notify chan error) *notifyErrorWaitGroupWrapper {
 	return &notifyErrorWaitGroupWrapper{
-		notify: notify,
-		cnt:    *atomic.NewUint64(0),
+		WaitGroupPool: util.NewWaitGroupPool(gp),
+		notify:        notify,
+		cnt:           *atomic.NewUint64(0),
 	}
 }
 

pkg/executor/builder.go

@@ -2890,11 +2890,14 @@ func (b *executorBuilder) buildAnalyzeColumnsPushdown(
 }
 
 func (b *executorBuilder) buildAnalyze(v *plannercore.Analyze) exec.Executor {
+	gp := domain.GetDomain(b.ctx).StatsHandle().GPool()
 	e := &AnalyzeExec{
 		BaseExecutor: exec.NewBaseExecutor(b.ctx, v.Schema(), v.ID()),
 		tasks:        make([]*analyzeTask, 0, len(v.ColTasks)+len(v.IdxTasks)),
 		opts:         v.Opts,
 		OptionsMap:   v.OptionsMap,
+		wg:           util.NewWaitGroupPool(gp),
+		gp:           gp,
 	}
 	autoAnalyze := ""
 	if b.ctx.GetSessionVars().InRestrictedSQL {

pkg/util/BUILD.bazel

@@ -43,6 +43,7 @@ go_library(
         "@com_github_pingcap_failpoint//:failpoint",
         "@com_github_pingcap_log//:log",
         "@com_github_pingcap_tipb//go-tipb",
+        "@com_github_tiancaiamao_gp//:gp",
         "@com_github_tikv_client_go_v2//oracle",
         "@io_etcd_go_etcd_client_v3//:client",
         "@io_etcd_go_etcd_client_v3//concurrency",

pkg/util/topsql/BUILD.bazel

@@ -28,6 +28,7 @@ go_test(
     ],
     embed = [":topsql"],
     flaky = True,
+    shard_count = 5,
     deps = [
         "//pkg/config",
         "//pkg/parser",

pkg/util/topsql/reporter/BUILD.bazel

@@ -45,6 +45,7 @@ go_test(
     ],
     embed = [":reporter"],
     flaky = True,
+    shard_count = 36,
     deps = [
         "//pkg/config",
         "//pkg/testkit/testsetup",

pkg/util/topsql/stmtstats/BUILD.bazel

@@ -28,6 +28,7 @@ go_test(
     ],
     embed = [":stmtstats"],
     flaky = True,
+    shard_count = 11,
     deps = [
         "//pkg/testkit/testsetup",
         "//pkg/util/topsql/state",

pkg/util/wait_group_wrapper.go

@@ -19,6 +19,7 @@ import (
 	"time"
 
 	"github.com/pingcap/tidb/pkg/util/logutil"
+	"github.com/tiancaiamao/gp"
 	"go.uber.org/zap"
 )
 
@@ -172,3 +173,27 @@ func (w *WaitGroupWrapper) RunWithRecover(exec func(), recoverFn func(r interfac
 		exec()
 	}()
 }
+
+// WaitGroupPool is a wrapper for sync.WaitGroup and support goroutine pool
+type WaitGroupPool struct {
+	sync.WaitGroup
+	gp *gp.Pool
+}
+
+// NewWaitGroupPool returns WaitGroupPool
+func NewWaitGroupPool(gp *gp.Pool) *WaitGroupPool {
+	var wg WaitGroupPool
+	wg.gp = gp
+	return &wg
+}
+
+// Run runs a function in a goroutine, adds 1 to WaitGroup
+// and calls done when function returns. Please DO NOT use panic
+// in the cb function.
+func (w *WaitGroupPool) Run(exec func()) {
+	w.Add(1)
+	w.gp.Go(func() {
+		defer w.Done()
+		exec()
+	})
+}

tools/tazel/util.go

@@ -45,5 +45,6 @@ func skipShardCount(path string) bool {
 		(strings.HasPrefix(path, "pkg/util") &&
 			!strings.HasPrefix(path, "pkg/util/admin") &&
 			!strings.HasPrefix(path, "pkg/util/chunk") &&
+			!strings.HasPrefix(path, "pkg/util/topsql") &&
 			!strings.HasPrefix(path, "pkg/util/stmtsummary"))
 }