[SyncFS] Don't re-run conflict resolution on resolution error

chrome/browser/sync_file_system/drive_backend/sync_engine.cc

Index: chrome/browser/sync_file_system/drive_backend/sync_engine.cc
diff --git a/chrome/browser/sync_file_system/drive_backend/sync_engine.cc b/chrome/browser/sync_file_system/drive_backend/sync_engine.cc
index 2b10199657e3d917a980d03f108cafb679524451..066aa8ae8482e97c570d6d3498a2718ae6d63cd7 100644
--- a/chrome/browser/sync_file_system/drive_backend/sync_engine.cc
+++ b/chrome/browser/sync_file_system/drive_backend/sync_engine.cc
@@ -570,6 +570,7 @@ void SyncEngine::MaybeStartFetchChanges() {
   base::TimeTicks now = base::TimeTicks::Now();
   if (!should_check_remote_change_ && now < time_to_check_changes_) {
     if (!metadata_database_->HasDirtyTracker() && should_check_conflict_) {
+      should_check_conflict_ = false;
       task_manager_->ScheduleSyncTaskIfIdle(
           scoped_ptr<SyncTask>(new ConflictResolver(this)),
           base::Bind(&SyncEngine::DidResolveConflict,
@@ -590,8 +591,8 @@ void SyncEngine::MaybeStartFetchChanges() {
 }
 
 void SyncEngine::DidResolveConflict(SyncStatusCode status) {

[nhiroki, 2014/01/30 04:24:35]

How about having "DCHECK(!should_check_conflict_);" here?

[nhiroki, 2014/01/30 04:53:27]

Hmm... this seems overkill...? Anyway, your change looks reasonable. lgtm

-  if (status == SYNC_STATUS_NO_CONFLICT)
-    should_check_conflict_ = false;
+  if (status == SYNC_STATUS_OK)
+    should_check_conflict_ = true;
 }
 
 void SyncEngine::DidFetchChanges(SyncStatusCode status) {