[turbofan] Instruction scheduler: keep ready nodes with same latency sorted in original order.

src/compiler/instruction-scheduler.cc

 // Copyright 2015 the V8 project authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "src/compiler/instruction-scheduler.h"
 
 #include "src/base/adapters.h"
 #include "src/base/utils/random-number-generator.h"
 
 namespace v8 {
 namespace internal {
 namespace compiler {
 
 void InstructionScheduler::SchedulingQueueBase::AddNode(
     ScheduleGraphNode* node) {
   // We keep the ready list sorted by total latency so that we can quickly find
   // the next best candidate to schedule.
   auto it = nodes_.begin();
   while ((it != nodes_.end()) &&
-         ((*it)->total_latency() > node->total_latency())) {
+         ((*it)->total_latency() >= node->total_latency())) {
     ++it;
   }
   nodes_.insert(it, node);
 }
 
 
 InstructionScheduler::ScheduleGraphNode*
 InstructionScheduler::CriticalPathFirstQueue::PopBestCandidate(int cycle) {
   DCHECK(!IsEmpty());
   auto candidate = nodes_.end();
@@ skipping 324 matching lines @@
       }
     }
 
     node->set_total_latency(max_latency + node->latency());
   }
 }
 
 }  // namespace compiler
 }  // namespace internal
 }  // namespace v8