diff --git a/core/parallel_state_processor.go b/core/parallel_state_processor.go
index 23a3f9b7e2..499f6231a6 100644
--- a/core/parallel_state_processor.go
+++ b/core/parallel_state_processor.go
@@ -543,23 +543,28 @@ func (p *ParallelStateProcessor) runSlotLoop(slotIndex int, slotType int32) {
 			nextIdx := p.mergedTxIndex.Load() + 1
 			if nextIdx < int32(len(p.allTxReqs)) {
 				nextMergeReq := p.allTxReqs[nextIdx]
-				if atomic.CompareAndSwapInt32(&nextMergeReq.runnable, 1, 0) {
-					// execute.
-					res := p.executeInSlot(slotIndex, nextMergeReq)
-					if res != nil {
-						p.txResultChan <- res
+				if nextMergeReq.runnable == 1 {
+					if atomic.CompareAndSwapInt32(&nextMergeReq.runnable, 1, 0) {
+						// execute.
+						res := p.executeInSlot(slotIndex, nextMergeReq)
+						if res != nil {
+							p.txResultChan <- res
+						}
 					}
 				}
 			}
-			// try the next req in loop sequence.
-			if !atomic.CompareAndSwapInt32(&txReq.runnable, 1, 0) {
-				continue
-			}
-			res := p.executeInSlot(slotIndex, txReq)
-			if res == nil {
-				continue
+
+			if txReq.runnable == 1 {
+				// try the next req in loop sequence.
+				if !atomic.CompareAndSwapInt32(&txReq.runnable, 1, 0) {
+					continue
+				}
+				res := p.executeInSlot(slotIndex, txReq)
+				if res == nil {
+					continue
+				}
+				p.txResultChan <- res
 			}
-			p.txResultChan <- res
 		}
 		// switched to the other slot.
 		if interrupted {
@@ -588,23 +593,27 @@ func (p *ParallelStateProcessor) runSlotLoop(slotIndex int, slotType int32) {
 			nextIdx := p.mergedTxIndex.Load() + 1
 			if nextIdx < int32(len(p.allTxReqs)) {
 				nextMergeReq := p.allTxReqs[nextIdx]
-				if atomic.CompareAndSwapInt32(&nextMergeReq.runnable, 1, 0) {
-					// execute.
-					res := p.executeInSlot(slotIndex, nextMergeReq)
-					if res != nil {
-						p.txResultChan <- res
+				if nextMergeReq.runnable == 1 {
+					if atomic.CompareAndSwapInt32(&nextMergeReq.runnable, 1, 0) {
+						// execute.
+						res := p.executeInSlot(slotIndex, nextMergeReq)
+						if res != nil {
+							p.txResultChan <- res
+						}
 					}
 				}
 			}
 
-			if !atomic.CompareAndSwapInt32(&stealTxReq.runnable, 1, 0) {
-				continue
-			}
-			res := p.executeInSlot(slotIndex, stealTxReq)
-			if res == nil {
-				continue
+			if stealTxReq.runnable == 1 {
+				if !atomic.CompareAndSwapInt32(&stealTxReq.runnable, 1, 0) {
+					continue
+				}
+				res := p.executeInSlot(slotIndex, stealTxReq)
+				if res == nil {
+					continue
+				}
+				p.txResultChan <- res
 			}
-			p.txResultChan <- res
 		}
 	}
 }
@@ -643,13 +652,15 @@ func (p *ParallelStateProcessor) runQuickMergeSlotLoop(slotIndex int, slotType i
 			if txReq.conflictIndex.Load() > p.mergedTxIndex.Load() {
 				break
 			}
-			if !atomic.CompareAndSwapInt32(&txReq.runnable, 1, 0) {
-				continue
-			}
-			res := p.executeInSlot(slotIndex, txReq)
-			if res != nil {
-				executed--
-				p.txResultChan <- res
+			if txReq.runnable == 1 {
+				if !atomic.CompareAndSwapInt32(&txReq.runnable, 1, 0) {
+					continue
+				}
+				res := p.executeInSlot(slotIndex, txReq)
+				if res != nil {
+					executed--
+					p.txResultChan <- res
+				}
 			}
 		}
 	}