1580618 - Perma Fission Tier 2 PROCESS-CRASH | Main app process exited normally | application crashed [@ nsPermissionManager::CommonTestPermissionInternal(nsIPrincipal*, nsIURI*, int, nsTSubstring<char> const&, unsigned int*, bool, bool)]

Status: RESOLVED FIXED

(Core :: DOM: Content Processes, defect, P2)

Description

[Treeherder Bug Filer]

Filed by: jwatt [at] jwatt.org
Parsed log: https://treeherder.mozilla.org/logviewer.html#?job_id=265999611&repo=try
Full log: https://queue.taskcluster.net/v1/task/BQVLfMQBRi2GJUE5Be1GuQ/runs/0/artifacts/public/logs/live_backing.log

---

The patch for bug 1559841 makes this perma fail for DEBUG Fission builds when running the test test_window_cross_origin_props.html due to:

Assertion failure: PermissionAvailable(prin, aType), at /builds/worker/workspace/build/src/extensions/permissions/nsPermissionManager.cpp:2386
dom/base/test/test_window_cross_origin_props.html | Should have focused first subframe - got "", expected "x"

As discussed with nika and kmag, we'll disable this test for now.

Comment 1

[Intermittent Failures Robot]

1 failures in 4637 pushes (0.0 failures/push) were associated with this bug in the last 7 days.

Repository breakdown:

• try: 1

Platform breakdown:

• linux64: 1

For more details, see:
https://treeherder.mozilla.org/intermittent-failures.html#/bugdetails?bug=1580618&startday=2019-09-09&endday=2019-09-15&tree=all

Comment 2

[BugBot [:suhaib / :marco/ :calixte]]

The priority flag is not set for this bug.
:jimm, could you have a look please?

For more information, please visit auto_nag documentation.

Comment 3

[Chris Peterson [:cpeterson]]

Henri, Hsin-Yi says you own this bug and that it depends on bug 1556627, a complex refactoring.

Comment 4

[Chris Peterson [:cpeterson]]

(In reply to Chris Peterson [:cpeterson] from comment #3)

Henri, Hsin-Yi says you own this bug and that it depends on bug 1556627, a complex refactoring.

SetFocus bug 1556627 has been fixed. Is additional work needed before we can try re-enabling this test (test_window_cross_origin_props.html)?

https://searchfox.org/mozilla-central/rev/2fd8ffcf087bc59a8e5c962965bbb7bf230bcd28/dom/base/test/mochitest.ini#843-845

Comment 5

[Henri Sivonen (:hsivonen)]

To avoid overlapping investigation, let's look at this after bug 1614268 has been fixed.

Comment 6

[Henri Sivonen (:hsivonen)]

(In reply to Henri Sivonen (:hsivonen) from comment #5)

To avoid overlapping investigation, let's look at this after bug 1614268 has been fixed.

Still fails with those patches applied.

Comment 7

[Henri Sivonen (:hsivonen)]

Possibly depends on bug 1613054.

Comment 8

[Henri Sivonen (:hsivonen)]

This indeed depends on bug 1613054. The test calls .focus() on an OOP iframe and wants to see it become focused synchronously. Normally, the focus code computes the element to focus within the iframe and then things propagate back up so that the iframe becomes the focused element of the DOM inner window that the iframe element is in.

When the iframe is OOP, there's no element to be found in it, and we don't have the synchronous propagation back up, either.

Comment 9

[Hsin-Yi Tsai (she/her) [:hsinyi]]

Hi Chris, I think we should move the Fission milestone to M6c per the dependency.

Comment 10

[Chris Peterson [:cpeterson]]

(In reply to Hsin-Yi Tsai [:hsinyi] from comment #9)

Hi Chris, I think we should move the Fission milestone to M6c per the dependency.

Sounds good since the dependency (bug 1613054) is an M6c bug.

Comment 11

[Henri Sivonen (:hsivonen)]

This is weird. I tried to create a WPT for the failure seen here in order to see what Chrome does: bug 1649099. However, the WPT I wrote already works in Fission. Yet, this test still fails.

Comment 12

[Neha Kochar [:neha]]

Anny, could you please re-test this with your about:blank patches? Nika thinks this might be fixed with your patches.

Comment 13

[Anny Gakhokidze [:annyG]]

My about:blank patches fix the crash, however, debug + fission configuration yields the following test failures

  FAIL Should have focused first subframe - got "", expected "x"
    SimpleTest.is@SimpleTest/SimpleTest.js:500:14
    window.onload@dom/base/test/test_window_cross_origin_props.html:26:7
  FAIL Should have focused second subframe - got "", expected "y"
    SimpleTest.is@SimpleTest/SimpleTest.js:500:14
    window.onload@dom/base/test/test_window_cross_origin_props.html:29:7
  FAIL Subframe is not closed - got true, expected false
    SimpleTest.is@SimpleTest/SimpleTest.js:500:14
    window.onload@dom/base/test/test_window_cross_origin_props.html:33:7
  FAIL Subframe is still not closed - got true, expected false
    SimpleTest.is@SimpleTest/SimpleTest.js:500:14
    window.onload@dom/base/test/test_window_cross_origin_props.html:35:7`

I read the comments here and in bug 1613054, and I'm not sure if the focus on OOP iframes is WIP or I should investigate the failures myself.

Comment 14

[Henri Sivonen (:hsivonen)]

Bug 1613054 is still a TODO. Currently, you should assume that calling .focus() on an OOP iframe element to be a little broken and calling .focus() on the .contentWindow of an OOP iframe to be very broken.

Comment 15

[Henri Sivonen (:hsivonen)]

BrowsingContext::Focus should adjust the activeElement of the embedder.

Comment 16

[Henri Sivonen (:hsivonen)]

This test was re-enabled by part 1 of bug 1613054.