2 files changed, 69 insertions, 24 deletions

diff --git a/app/services/git/base_hooks_service.rb b/app/services/git/base_hooks_service.rb
index d30df34e54b..47c308c8280 100644
--- a/app/services/git/base_hooks_service.rb
+++ b/app/services/git/base_hooks_service.rb
@@ -8,8 +8,6 @@ module Git
     PROCESS_COMMIT_LIMIT = 100
 
     def execute
-      project.repository.after_create if project.empty_repo?
-
       create_events
       create_pipelines
       execute_project_hooks
@@ -19,7 +17,7 @@ module Git
 
       update_remote_mirrors
 
-      push_data
+      success
     end
 
     private
@@ -33,7 +31,7 @@ module Git
     end
 
     def limited_commits
-      commits.last(PROCESS_COMMIT_LIMIT)
+      @limited_commits ||= commits.last(PROCESS_COMMIT_LIMIT)
     end
 
     def commits_count
@@ -48,43 +46,69 @@ module Git
       []
     end
 
+    # Push events in the activity feed only show information for the
+    # last commit.
     def create_events
-      EventCreateService.new.push(project, current_user, push_data)
+      EventCreateService.new.push(project, current_user, event_push_data)
     end
 
     def create_pipelines
       return unless params.fetch(:create_pipelines, true)
 
       Ci::CreatePipelineService
-        .new(project, current_user, push_data)
+        .new(project, current_user, pipeline_params)
         .execute(:push, pipeline_options)
     end
 
     def execute_project_hooks
-      project.execute_hooks(push_data, hook_name)
-      project.execute_services(push_data, hook_name)
+      # Creating push_data invokes one CommitDelta RPC per commit. Only
+      # build this data if we actually need it.
+      project.execute_hooks(push_data, hook_name) if project.has_active_hooks?(hook_name)
+      project.execute_services(push_data, hook_name) if project.has_active_services?(hook_name)
     end
 
     def enqueue_invalidate_cache
-      ProjectCacheWorker.perform_async(
-        project.id,
-        invalidated_file_types,
-        [:commit_count, :repository_size]
-      )
+      file_types = invalidated_file_types
+
+      return unless file_types.present?
+
+      ProjectCacheWorker.perform_async(project.id, file_types, [], false)
     end
 
-    def push_data
-      @push_data ||= Gitlab::DataBuilder::Push.build(
-        project: project,
-        user: current_user,
+    def pipeline_params
+      {
+        before: params[:oldrev],
+        after: params[:newrev],
+        ref: params[:ref],
+        push_options: params[:push_options] || {},
+        checkout_sha: Gitlab::DataBuilder::Push.checkout_sha(
+          project.repository, params[:newrev], params[:ref])
+      }
+    end
+
+    def push_data_params(commits:, with_changed_files: true)
+      {
         oldrev: params[:oldrev],
         newrev: params[:newrev],
         ref: params[:ref],
-        commits: limited_commits,
+        project: project,
+        user: current_user,
+        commits: commits,
         message: event_message,
         commits_count: commits_count,
-        push_options: params[:push_options] || {}
-      )
+        with_changed_files: with_changed_files
+      }
+    end
+
+    def event_push_data
+      # We only need the last commit for the event push, and we don't
+      # need the full deltas either.
+      @event_push_data ||= Gitlab::DataBuilder::Push.build(
+        push_data_params(commits: commits.last, with_changed_files: false))
+    end
+
+    def push_data
+      @push_data ||= Gitlab::DataBuilder::Push.build(push_data_params(commits: limited_commits))
 
       # Dependent code may modify the push data, so return a duplicate each time
       @push_data.dup
diff --git a/app/services/git/branch_hooks_service.rb b/app/services/git/branch_hooks_service.rb
index c41f445c3c4..d2b037a680c 100644
--- a/app/services/git/branch_hooks_service.rb
+++ b/app/services/git/branch_hooks_service.rb
@@ -63,7 +63,7 @@ module Git
     end
 
     def branch_create_hooks
-      project.repository.after_create_branch
+      project.repository.after_create_branch(expire_cache: false)
       project.after_create_default_branch if default_branch?
     end
 
@@ -78,13 +78,21 @@ module Git
     end
 
     def branch_remove_hooks
-      project.repository.after_remove_branch
+      project.repository.after_remove_branch(expire_cache: false)
     end
 
     # Schedules processing of commit messages
     def enqueue_process_commit_messages
-      limited_commits.each do |commit|
-        next unless commit.matches_cross_reference_regex?
+      referencing_commits = limited_commits.select(&:matches_cross_reference_regex?)
+
+      upstream_commit_ids = upstream_commit_ids(referencing_commits)
+
+      referencing_commits.each do |commit|
+        # Avoid reprocessing commits that already exist upstream if the project
+        # is a fork. This will prevent duplicated/superfluous system notes on
+        # mentionables referenced by a commit that is pushed to the upstream,
+        # that is then also pushed to forks when these get synced by users.
+        next if upstream_commit_ids.include?(commit.id)
 
         ProcessCommitWorker.perform_async(
           project.id,
@@ -142,5 +150,18 @@ module Git
     def branch_name
       strong_memoize(:branch_name) { Gitlab::Git.ref_name(params[:ref]) }
     end
+
+    def upstream_commit_ids(commits)
+      set = Set.new
+
+      upstream_project = project.fork_source
+      if upstream_project
+        upstream_project
+          .commits_by(oids: commits.map(&:id))
+          .each { |commit| set << commit.id }
+      end
+
+      set
+    end
   end
 end