The first difference (lack of "line 10: " with OOPIFs) is kind of WAI.

I believe the second difference indicates that updateLayoutIgnorePendingStylesheetsAndRunPostLayoutTasks doesn't work with remote frames (aka OOPIFs - out of process iframes):

third_party/WebKit/Source/core/testing/Internals.cpp:

void Internals::updateLayoutIgnorePendingStylesheetsAndRunPostLayoutTasks(
    Node* node,
    ExceptionState& exceptionState) {
  Document* document = nullptr;
  if (!node) {
    document = contextDocument();
  } else if (node->isDocumentNode()) {
    document = toDocument(node);
  } else if (isHTMLIFrameElement(*node)) {
    // WE HAVE A PROBLEM HERE?  AFAIK contentDocument will be null for an OOPIF.
    document = toHTMLIFrameElement(*node).contentDocument();
  }

  if (!document) {
    exceptionState.throwTypeError(
        "The node provided is neither a document nor an IFrame.");
    return;
  }
  document->updateStyleAndLayoutIgnorePendingStylesheets(
      Document::RunPostLayoutTasksSynchronously);
}

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/0c709a831773d24b82e64888eae63e3dda9c9d66

commit 0c709a831773d24b82e64888eae63e3dda9c9d66
Author: lukasza <lukasza@chromium.org>
Date: Mon Oct 31 16:59:12 2016

Updating --site-per-process expectations for insecure-plugin-in-iframe.html

The test stopped crashing in --site-per-process mode after r428312.
A new bug has been opened for the remaining text diff in test output.

BUG= 582494 ,  660872 
NOTRY=true

Review-Url: https://codereview.chromium.org/2466593002
Cr-Commit-Position: refs/heads/master@{#428745}

[modify] https://crrev.com/0c709a831773d24b82e64888eae63e3dda9c9d66/third_party/WebKit/LayoutTests/FlagExpectations/site-per-process

I don't understand why we'd need to call updateLayoutIgnorePendingStylesheetsAndRunPostLayoutTasks() explicitly in the test. It's done on begin-frame regardlessly? Isn't that enough?

tkent@ - would you remember why you've added the call to updateLayoutIgnorePendingStylesheetsAndRunPostLayoutTasks in third_party/WebKit/LayoutTests/http/tests/security/mixedContent/insecure-plugin-in-iframe.html in https://codereview.chromium.org/61763016?  As rune@ points out it does seem unneeded (i.e. the mixed content message appears even without this call).  Would it be okay to just remove the call from this particular test?

Comment 5: It was necessary to load a plugin in the IFRAME immediately.  I'm not sure if it is unnecessary now.  At worst, removing it might make the test flaky.

lukasza@: If you wrap the internals call in a try/catch, you will run the updateLayoutIgnore... in the LocalFrame case, and get the same expected output for the OOPIF case (it won't have an effect regardlessly).

For http/tests/security/mixedContent/insecure-plugin-in-iframe.html, I think I'll just move the call to updateLayoutIgnorePendingStylesheetsAndRunPostLayoutTasks from the parent into the child frame (so that the call always happens on a local document).  This is a cowardly^H^H^Hlow-risk CL that will hopefully pave way toward enabling this test with --site-per-process (although I agree that in principle we might want to investigate getting rid of that call altogether).

Once the above is done, I can try to dust off the CL (https://codereview.chromium.org/2191143002) that allows layout tests to strip/ignore console line numbers (tracked broadly by  issue 619662 ).

Fix proposed at https://codereview.chromium.org/2468953002
(it turns out that console line numbers can be tackled without going back to https://codereview.chromium.org/2191143002 for help).

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/cd8d931dc59e273273701feae744adff8714c27f

commit cd8d931dc59e273273701feae744adff8714c27f
Author: lukasza <lukasza@chromium.org>
Date: Tue Nov 01 23:15:21 2016

Make mixedContent/insecure-plugin-in-iframe.html test OOPIF-compatible.

The CL makes the following changes:

- Moves the call to
  internals.updateLayoutIgnorePendingStylesheetsAndRunPostLayoutTasks
  from the parent frame into the child frame.  This ensures that this
  method works on a local document - this is important because the
  method doesn't support OOPIFs.

- Delays the call to
  internals.updateLayoutIgnorePendingStylesheetsAndRunPostLayoutTasks
  via window.setTimeout.  This ensures that the mixed content message
  does not include the line number of the call (and so appears the same
  with an without OOPIFs).  The line number used to indicate which line
  called internals.updateLayoutIgnore...heetsAndRunPostLayoutTasks and
  this piece of information is artificial (and not essential to test
  for) - no real page will call this test-only window.internals method.

- To make sure the test finishes *after* the delayed call to
  internals.updateLayoutIgnorePendingStylesheetsAndRunPostLayoutTasks
  a pair of testRunner.waitUntilDone / testRunner.notifyDone calls
  has been added.  Note that the subframe used to post a "done"
  message back to the opener, but this message was never listened to
  (and so the postMessage call can be deleted).

BUG= 660872 

Review-Url: https://codereview.chromium.org/2468953002
Cr-Commit-Position: refs/heads/master@{#429147}

[modify] https://crrev.com/cd8d931dc59e273273701feae744adff8714c27f/third_party/WebKit/LayoutTests/FlagExpectations/site-per-process
[modify] https://crrev.com/cd8d931dc59e273273701feae744adff8714c27f/third_party/WebKit/LayoutTests/http/tests/security/mixedContent/insecure-plugin-in-iframe-expected.txt
[modify] https://crrev.com/cd8d931dc59e273273701feae744adff8714c27f/third_party/WebKit/LayoutTests/http/tests/security/mixedContent/insecure-plugin-in-iframe.html
[modify] https://crrev.com/cd8d931dc59e273273701feae744adff8714c27f/third_party/WebKit/LayoutTests/http/tests/security/mixedContent/resources/frame-with-insecure-plugin.html

Hmmm... it seems that http/tests/security/mixedContent/insecure-plugin-in-iframe.html runs as expected now.  I am guessing that the CL from #c10 above fixed the problem and I just forgot to mark the bug as fixed.