PlzNavigate: implement process reuse for ServiceWorkers

content/browser/frame_host/render_frame_host_manager.cc

Index: content/browser/frame_host/render_frame_host_manager.cc
diff --git a/content/browser/frame_host/render_frame_host_manager.cc b/content/browser/frame_host/render_frame_host_manager.cc
index 8ab2e551bde024f90f81605062548a4146eab23b..131fe0ae9fe1e239c6bad6d0898b08cdc729117d 100644
--- a/content/browser/frame_host/render_frame_host_manager.cc
+++ b/content/browser/frame_host/render_frame_host_manager.cc
@@ -484,7 +484,6 @@ void RenderFrameHostManager::CommitPendingIfNecessary(
     // https://code.google.com/p/chromium/issues/detail?id=75195
     if (was_caused_by_user_gesture) {
       if (IsBrowserSideNavigationEnabled()) {

[Charlie Reis, 2017/05/24 17:18:37]

Did this change cause
NavigatorTestWithBrowserSideNavigation.SimpleRendererInitiatedCrossSiteNavigation
to fail in --site-per-process plus PlzNavigate?

https://build.chromium.org/p/tryserver.chromium.linux/builders/linux_site_iso...

(I wonder if that's just a test expectation question or if it represents a real
issue?)

[clamy, 2017/05/24 18:03:04]

NavigatorTestWithBrowserSideNavigation.SimpleRendererInitiatedCrossSiteNavigation
was wrong for --site-per-process and this exposed it. This has now been
corrected.

However, this change was not the right one: the call to CleanUpNavigation
shouldn't be removed but put after the call to ResetNavigationRequest (this made
other tests fail with PlzNavigate). All should be good now.

-        CleanUpNavigation();
         frame_tree_node_->ResetNavigationRequest(false, true);
       } else {
         CancelPending();