Fix embedding token failure under CreateNewHostForSameSiteSubframe.

With CreateNewHostForSameSiteSubframe enabled, a new frame is created
and so we expect the embedding token to change.

Bug: 1064944
Change-Id: I8f9833cc5942c847b78a93ddfa021597771f5c34
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2138991Reviewed-by: Arthur Sonzogni <arthursonzogni@chromium.org>
Reviewed-by: Calder Kitagawa <ckitagawa@chromium.org>
Commit-Queue: Fergal Daly <fergal@chromium.org>
Cr-Commit-Position: refs/heads/master@{#757062}

content/browser/frame_host/embedding_token_browsertest.cc

@@ -5,6 +5,7 @@
 #include "base/unguessable_token.h"
 #include "content/browser/frame_host/frame_tree.h"
 #include "content/browser/web_contents/web_contents_impl.h"
+#include "content/common/content_navigation_policy.h"
 #include "content/public/test/browser_test_utils.h"
 #include "content/public/test/content_browser_test.h"
 #include "content/public/test/content_browser_test_utils.h"
@@ -169,7 +170,13 @@ IN_PROC_BROWSER_TEST_F(EmbeddingTokenBrowserTest,
                       b_url.Resolve("valid.html"));
   auto new_child_0_token = root->child_at(0)->GetEmbeddingToken();
   ASSERT_TRUE(new_child_0_token.has_value());
-  EXPECT_EQ(child_0_token, new_child_0_token);
+  // If we are creating a new frame even for same-site navigations then we would
+  // expect a different token.
+  if (CreateNewHostForSameSiteSubframe()) {
+    EXPECT_NE(child_0_token, new_child_0_token);
+  } else {
+    EXPECT_EQ(child_0_token, new_child_0_token);
+  }
 
   // TODO(ckitagawa): Somehow assert that the parent and child have matching
   // embedding tokens in parent HTMLOwnerElement and child LocalFrame.