PlzNavigate: Improvements to RFHM commit logic.

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 29771b2ab65da73c748a60643a0634acce2ec5a2..207bcc0edc112d60dc4a22ac91e05fcf1e571c27 100644
--- a/content/browser/frame_host/render_frame_host_manager.cc
+++ b/content/browser/frame_host/render_frame_host_manager.cc
@@ -460,9 +460,14 @@ void RenderFrameHostManager::CommitPendingIfNecessary(
     if (render_frame_host == speculative_render_frame_host_.get()) {
       CommitPending();
     } else if (render_frame_host == render_frame_host_.get()) {
-      // TODO(carlosk): this code doesn't properly handle in-page navigation or
-      // interwoven navigation requests.
-      DCHECK(!speculative_render_frame_host_);
+      DCHECK(!should_reuse_web_ui_ || web_ui_);
+      // TODO(carlosk): this code might not handle interwoven navigation

[clamy, 2015/03/31 11:40:42]

Could you add a comment explaining that even if there is no speculative RFH
there may be a speculative webui to commit?

[carlosk, 2015/03/31 14:19:39]

Done.

+      // requests properly.

[carlosk, 2015/03/27 17:11:29]

Due to the possibility of having multiple ongoing navigations and the
asynchronous nature of message passing there still are situations where the code
below won't do the right thing.

[clamy, 2015/03/31 11:40:42]

Yes we should maybe think about pausing a navigation request when we are in the
process of committing something just to make things clearer.

[carlosk, 2015/03/31 14:19:39]

Let's talk offline about this as I don't understand what you meant by "pausing a
navigation".

+      if (speculative_web_ui_ && !speculative_render_frame_host_) {
+        CommitPending();
+      } else {
+        CleanUpNavigation();
+      }
     } else {
       // No one else should be sending us a DidNavigate in this state.
       DCHECK(false);
@@ -1637,8 +1642,14 @@ void RenderFrameHostManager::CommitPending() {
     }
   } else {
     // PlzNavigate
-    if (!should_reuse_web_ui_)
+    if (speculative_web_ui_) {
+      DCHECK(!should_reuse_web_ui_);
       web_ui_.reset(speculative_web_ui_.release());
+    } else if (should_reuse_web_ui_) {

[clamy, 2015/03/31 11:40:42]

At which point is should_reuse_web_ui_ reset?

[carlosk, 2015/03/31 14:19:39]

You are right to bring this up: this is the place.

+      CHECK(web_ui_);

[clamy, 2015/03/31 11:40:42]

I wonder if this should be a DCHECK seeing that all other webui related checks
are DCHECKs instead of CHECKs.

[carlosk, 2015/03/31 14:19:39]

Done.

+    } else {
+      web_ui_.reset();
+    }

[carlosk, 2015/03/27 17:11:30]

This code does basically the same as the previous one but looks way clearer. I
was really confused until I found out that speculative_web_ui_.release() won't
raise an exception if it points to NULL (and correctly caused the reset of
|web_ui_|).

     DCHECK(!speculative_web_ui_);
   }