1930090 - Retry dispatching single event if navigation aborted a dispatch

Status: RESOLVED FIXED

(Remote Protocol :: WebDriver BiDi, defect, P2)

Description

[Treeherder Bug Filer]

Filed by: hskupin [at] mozilla.com
Parsed log: https://treeherder.mozilla.org/logviewer?job_id=480944459&repo=mozilla-central
Full log: https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/SAQZV7MPT-C2L_dQnLTzhw/runs/0/artifacts/public/logs/live_backing.log

---

Comment 1

[Henrik Skupin [:whimboo][⌚️UTC+2]]

This is a new test as introduced with my work on bug 1921314. It shows that when a navigation happens while we are in the middle of sending the mouse event through EventUtils we raise an error like:

[task 2024-11-03T10:03:32.578Z] 10:03:32     INFO - E       webdriver.bidi.error.UnknownErrorException: unknown error ([Exception... "Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIDOMWindowUtils.sendMouseEvent]"  nsresult: "0x80004005 (NS_ERROR_FAILURE)"  location: "JS frame :: chrome://remote/content/external/EventUtils.js :: synthesizeMouseAtPoint :: line 788"  data: no])
[task 2024-11-03T10:03:32.578Z] 10:03:32     INFO - E       
[task 2024-11-03T10:03:32.578Z] 10:03:32     INFO - E       Remote-end stacktrace:
[task 2024-11-03T10:03:32.578Z] 10:03:32     INFO - E       
[task 2024-11-03T10:03:32.578Z] 10:03:32     INFO - E       synthesizeMouseAtPoint@chrome://remote/content/external/EventUtils.js:788:32
[task 2024-11-03T10:03:32.578Z] 10:03:32     INFO - E       event.synthesizeMouseAtPoint@chrome://remote/content/shared/webdriver/Event.sys.mjs:83:30
[task 2024-11-03T10:03:32.578Z] 10:03:32     INFO - E       _dispatchEvent@chrome://remote/content/webdriver-bidi/modules/windowglobal/input.sys.mjs:154:20
[task 2024-11-03T10:03:32.578Z] 10:03:32     INFO - E       handleCommand@chrome://remote/content/shared/messagehandler/MessageHandler.sys.mjs:257:33
[task 2024-11-03T10:03:32.578Z] 10:03:32     INFO - E       receiveMessage@chrome://remote/content/shared/messagehandler/transports/js-window-actors/MessageHandlerFrameChild.sys.mjs:78:37
[task 2024-11-03T10:03:32.578Z] 10:03:32     INFO - 
[task 2024-11-03T10:03:32.578Z] 10:03:32     INFO - cmd_name   = 'input.performActions'
[task 2024-11-03T10:03:32.578Z] 10:03:32     INFO - future     = <Future finished exception=UnknownErrorException(unknown error, [Exception... "Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIDOMWindowUtils.sendMouseEvent]"  nsresult: "0x80004005 (NS_ERROR_FAILURE)"  location: "JS frame :: chrome://remote/content/external/EventUtils.js :: synthesizeMouseAtPoint :: line 788"  data: no], synthesizeMouseAtPoint@chrome://remote/content/external/EventUtils.js:788:32
[task 2024-11-03T10:03:32.578Z] 10:03:32     INFO - event.synthesizeMouseAtPoint@chrome://remote/content/shared/webdriver/Event.sys.mjs:83:30
[task 2024-11-03T10:03:32.579Z] 10:03:32     INFO - _dispatchEvent@chrome://remote/content/webdriver-bidi/modules/windowglobal/input.sys.mjs:154:20
[task 2024-11-03T10:03:32.579Z] 10:03:32     INFO - handleCommand@chrome://remote/content/shared/messagehandler/MessageHandler.sys.mjs:257:33
[task 2024-11-03T10:03:32.579Z] 10:03:32     INFO - receiveMessage@chrome://remote/content/shared/messagehandler/transports/js-window-actors/MessageHandlerFrameChild.sys.mjs:78:37
[task 2024-11-03T10:03:32.579Z] 10:03:32     INFO - )>

We as well have quite a few meta data entries for this test:
https://searchfox.org/mozilla-central/source/testing/web-platform/meta/webdriver/tests/bidi/input/perform_actions/navigation.py.ini

Comment 2

[Henrik Skupin [:whimboo][⌚️UTC+2]]

There are as well upstream failures with Chrome. Given the results the pointer action on the new page hasn't been sent.

I need to further investigate.

Comment 3

[Henrik Skupin [:whimboo][⌚️UTC+2]]

For this particular issue we are not getting a clear error type from the platform code, but I think that based on a NS_ERROR_FAILURE we should consider to just retry the command because it is dispatching a single event only and it was not successful.

To accomplish that we could throw an AbortError exception so that the RootTransport implementation understands it as a JSWindowActor disconnect and would retry the command a given number of times. That would allow us to continue the action sequence while in the other hand will finally throw that error if it is permanent.

Comment 4

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Attachment: Bug 1930090 - [webdriver-bidi] Retry dispatching single event if navigation aborted a dispatch.

Comment 5

[Intermittent Failures Robot]

3 failures in 4515 pushes (0.001 failures/push) were associated with this bug in the last 7 days.

Repository breakdown:

• autoland: 3

Platform and build breakdown:

• linux1804-64-qr: 2
  • debug: 2
• android-em-7-0-x86_64-qr: 1
  • debug: 1

For more details, see:
https://treeherder.mozilla.org/intermittent-failures/bugdetails?bug=1930090&startday=2024-11-04&endday=2024-11-10&tree=all

Comment 6

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/e96c75cc6b8e
[webdriver-bidi] Retry dispatching single event if navigation aborted a dispatch. r=webdriver-reviewers,Sasha

Comment 7

[Serban Stanca [:SerbanS]]

https://hg.mozilla.org/mozilla-central/rev/e96c75cc6b8e