Closed Bug 1378577 Opened 2 years ago Closed 2 years ago

Intermittent stylo layout/style/test/chrome/test_stylesheet_clone_import_rule.html | Test timed out.

Categories

(Core :: CSS Parsing and Computation, defect, P4)

defect

Tracking

()

RESOLVED FIXED
mozilla57
Tracking Status
firefox56 --- fixed
firefox57 --- fixed

People

(Reporter: intermittent-bug-filer, Assigned: bradwerth)

References

Details

(Keywords: intermittent-failure, Whiteboard: [stockwell fixed:race])

Attachments

(1 file)

Priority: -- → P4
No longer blocks: stylo-bustage
This failure picked up in frequency on August 15th, primarily for linux64-stylo-opt (36 failures in the last 7 days).

Here is a recent log:
https://treeherder.mozilla.org/logviewer.html#?repo=autoland&job_id=124674692&lineNumber=2737

and a related screenshot:
https://public-artifacts.taskcluster.net/f9Ms3Vc9R-KDRv4vPcmy1w/0/public/test_info//mozilla-test-fail-screenshot_FAEywg.png

here is data from the above log showing the failure:
[task 2017-08-21T21:11:19.132865Z] 21:11:19     INFO - TEST-START | layout/style/test/chrome/test_stylesheet_clone_import_rule.html
[task 2017-08-21T21:16:46.979622Z] 21:16:46     INFO - TEST-INFO | started process screentopng
[task 2017-08-21T21:16:47.263028Z] 21:16:47     INFO - TEST-INFO | screentopng: exit 0
[task 2017-08-21T21:16:47.264665Z] 21:16:47     INFO - Buffered messages logged at 21:11:19
[task 2017-08-21T21:16:47.266544Z] 21:16:47     INFO - TEST-PASS | layout/style/test/chrome/test_stylesheet_clone_import_rule.html | div begins as green. 
[task 2017-08-21T21:16:47.267358Z] 21:16:47     INFO - TEST-PASS | layout/style/test/chrome/test_stylesheet_clone_import_rule.html | Imported sheets exist. 
[task 2017-08-21T21:16:47.268015Z] 21:16:47     INFO - Buffered messages finished
[task 2017-08-21T21:16:47.268778Z] 21:16:47     INFO - TEST-UNEXPECTED-FAIL | layout/style/test/chrome/test_stylesheet_clone_import_rule.html | Test timed out. 
[task 2017-08-21T21:16:47.269555Z] 21:16:47     INFO - reportError@chrome://mochikit/content/tests/SimpleTest/TestRunner.js:121:7
[task 2017-08-21T21:16:47.269744Z] 21:16:47     INFO - TestRunner._checkForHangs@chrome://mochikit/content/tests/SimpleTest/TestRunner.js:142:7
[task 2017-08-21T21:16:47.269918Z] 21:16:47     INFO - setTimeout handler*TestRunner._checkForHangs@chrome://mochikit/content/tests/SimpleTest/TestRunner.js:163:5
[task 2017-08-21T21:16:47.270899Z] 21:16:47     INFO - setTimeout handler*TestRunner._checkForHangs@chrome://mochikit/content/tests/SimpleTest/TestRunner.js:163:5
[task 2017-08-21T21:16:47.271105Z] 21:16:47     INFO - setTimeout handler*TestRunner._checkForHangs@chrome://mochikit/content/tests/SimpleTest/TestRunner.js:163:5
[task 2017-08-21T21:16:47.271916Z] 21:16:47     INFO - setTimeout handler*TestRunner._checkForHangs@chrome://mochikit/content/tests/SimpleTest/TestRunner.js:163:5
[task 2017-08-21T21:16:47.272653Z] 21:16:47     INFO - setTimeout handler*TestRunner._checkForHangs@chrome://mochikit/content/tests/SimpleTest/TestRunner.js:163:5
[task 2017-08-21T21:16:47.272856Z] 21:16:47     INFO - setTimeout handler*TestRunner._checkForHangs@chrome://mochikit/content/tests/SimpleTest/TestRunner.js:163:5
[task 2017-08-21T21:16:47.273512Z] 21:16:47     INFO - setTimeout handler*TestRunner._checkForHangs@chrome://mochikit/content/tests/SimpleTest/TestRunner.js:163:5
[task 2017-08-21T21:16:47.274321Z] 21:16:47     INFO - setTimeout handler*TestRunner._checkForHangs@chrome://mochikit/content/tests/SimpleTest/TestRunner.js:163:5


:jet, I see you are the triage owner for this component- is this something you can help find someone to look into this sometime in the next 2 weeks?
Flags: needinfo?(bugs)
Whiteboard: [stockwell needswork]
:emilio, maybe you have ideas about this @import test issue, or can redirect to someone?
Flags: needinfo?(emilio+bugs)
Bug 1387933 was just fixed, which might be related to this, not sure.
It may indeed be related to that, but I'm suspicious about the test. In particular:

    stylesheet.insertRule('@import url("import_useless2.css")', 0);
    stylesheet.insertRule('@import url("import_useless2.css")', 1);

    // Wait for two StyleRuleAdded events to be fired.
    await ContentTaskUtils.waitForEvent(document, "StyleRuleAdded", true);
    await ContentTaskUtils.waitForEvent(document, "StyleRuleAdded", true);

What prevents one of the two events from fire before we call waitForEvent? I'm not familiar with ContentTaskUtils at all, but that looks slightly fishy to me...
Flags: needinfo?(emilio+bugs)
(In reply to Emilio Cobos Álvarez [:emilio] from comment #14)
> It may indeed be related to that, but I'm suspicious about the test. In
> particular:
> 
>     stylesheet.insertRule('@import url("import_useless2.css")', 0);
>     stylesheet.insertRule('@import url("import_useless2.css")', 1);
> 
>     // Wait for two StyleRuleAdded events to be fired.
>     await ContentTaskUtils.waitForEvent(document, "StyleRuleAdded", true);
>     await ContentTaskUtils.waitForEvent(document, "StyleRuleAdded", true);
> 
> What prevents one of the two events from fire before we call waitForEvent?
> I'm not familiar with ContentTaskUtils at all, but that looks slightly fishy
> to me...

You're right. I missed that while constructing this test. I find a way to fix it while preserving the intent.
Assignee: nobody → bwerth
Patch pushed.
Flags: needinfo?(bugs)
Attachment #8902042 - Flags: review?(emilio)
See Also: → 1394632
Comment on attachment 8902042 [details]
Bug 1378577: Fix await race condition which causes intermittent timeouts in test_stylesheet_clone_import_rule.html.

https://reviewboard.mozilla.org/r/173448/#review178818

Nice solution! I didn't know about the `checkFn` argument, that's why I wasn't aware of any trivial solution to this (which is why I didn't post a patch in the first place).

r=me, with the style issues double-checked with the style guide, or someone that writes more Firefox JS than I :)

::: layout/style/test/chrome/test_stylesheet_clone_import_rule.html:42
(Diff revision 1)
> +    let generateCountingListenerCallback = function(numberOfEventsToWaitFor) {
> +      let eventsReceived = 0;
> +      return function(event) {
> +        return (++eventsReceived >= numberOfEventsToWaitFor);
> +      }
> +    }

nit: Not sure if you're supposed to put a semicolon here or not, please double-check. As someone used to rust, it looks slightly clearer to me, but not sure what the JS style guide is.

::: layout/style/test/chrome/test_stylesheet_clone_import_rule.html:43
(Diff revision 1)
> +      let eventsReceived = 0;
> +      return function(event) {
> +        return (++eventsReceived >= numberOfEventsToWaitFor);
> +      }
> +    }
> +    let gotAllOurStyleRuleAddedEvents = ContentTaskUtils.waitForEvent(document,

nit: Indentation here looks somewhat off (though I'm not sure what the preferred style is for javascript).
Attachment #8902042 - Flags: review?(emilio) → review+
Comment on attachment 8902042 [details]
Bug 1378577: Fix await race condition which causes intermittent timeouts in test_stylesheet_clone_import_rule.html.

https://reviewboard.mozilla.org/r/173448/#review178820

::: layout/style/test/chrome/test_stylesheet_clone_import_rule.html:42
(Diff revision 1)
> +    let generateCountingListenerCallback = function(numberOfEventsToWaitFor) {
> +      let eventsReceived = 0;
> +      return function(event) {
> +        return (++eventsReceived >= numberOfEventsToWaitFor);
> +      }
> +    }

Yes, I'd suggest a semicolon here.  (We have ESLint to check JS style, but it's not enabled for the layout directory.)

::: layout/style/test/chrome/test_stylesheet_clone_import_rule.html:43
(Diff revision 1)
> +      let eventsReceived = 0;
> +      return function(event) {
> +        return (++eventsReceived >= numberOfEventsToWaitFor);
> +      }
> +    }
> +    let gotAllOurStyleRuleAddedEvents = ContentTaskUtils.waitForEvent(document,

Not sure there's one style for this at the moment...  This is probably okay!
Pushed by bwerth@mozilla.com:
https://hg.mozilla.org/integration/autoland/rev/7f2bcb7a902d
Fix await race condition which causes intermittent timeouts in test_stylesheet_clone_import_rule.html. r=emilio
https://hg.mozilla.org/mozilla-central/rev/7f2bcb7a902d
Status: NEW → RESOLVED
Closed: 2 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla57
https://hg.mozilla.org/releases/mozilla-beta/rev/f105366cc45d
Whiteboard: [stockwell needswork] → [stockwell fixed]
Should we reopen this bug or open another bug?

I guess I cannot land bug 1387993 with this intermittent there, because it seems to be high frequent. I guess I need to either disable that test or wait for this intermittent to be completely fixed.
Flags: needinfo?(bwerth)
The error log is very weird. The last test-pass message we get is from http://searchfox.org/mozilla-central/rev/51b3d67a5ec1758bd2fe7d7b6e75ad6b6b5da223/layout/style/test/chrome/test_stylesheet_clone_import_rule.html#47. The test then times out before we get to the next test-pass a few lines down at http://searchfox.org/mozilla-central/rev/51b3d67a5ec1758bd2fe7d7b6e75ad6b6b5da223/layout/style/test/chrome/test_stylesheet_clone_import_rule.html#53. If the log is accurate, then that would seem to indicate that the test is hanging on retrieving the "cssRules" property off a stylesheet DOM object.

I'll see if I can figure out what's going on. I'll open a new bug.
Flags: needinfo?(bwerth)
Blocks: 1395322
Whiteboard: [stockwell fixed] → [stockwell fixed:race]
You need to log in before you can comment on or make changes to this bug.