| Summary: | Regression(r236512): http/tests/navigation/keyboard-events-during-provisional-navigation.html is flaky | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | WebKit | Reporter: | Chris Dumez <cdumez> | ||||||||||||||
| Component: | WebKit2 | Assignee: | Chris Dumez <cdumez> | ||||||||||||||
| Status: | RESOLVED FIXED | ||||||||||||||||
| Severity: | Normal | CC: | achristensen, commit-queue, dbates, ews-watchlist, ggaren, rniwa, simon.fraser, tsavell, webkit-bug-importer | ||||||||||||||
| Priority: | P1 | Keywords: | InRadar, Regression | ||||||||||||||
| Version: | WebKit Nightly Build | ||||||||||||||||
| Hardware: | Unspecified | ||||||||||||||||
| OS: | Unspecified | ||||||||||||||||
| Bug Depends on: | 190112 | ||||||||||||||||
| Bug Blocks: | 189927 | ||||||||||||||||
| Attachments: |
|
||||||||||||||||
|
Description
Chris Dumez
2018-09-27 13:38:16 PDT
The test does:
console.log("Pressing \"z\" with access key modifiers should navigate to resources/keyboard-events-after-navigation.html.");
eventSender.keyDown("z", internals.accessKeyModifiers());
The console logging does an async IPC to the UIProcess and the eventSender.keyDown() does a sync IPC to the UIProcess. As a result, the execution order of those 2 is not pre-determined.
I suspect this has something to do with the IPC::SendSyncOption::UseFullySynchronousModeForTesting IPC send option that is used by WK2 eventSender. (In reply to Chris Dumez from comment #2) > I suspect this has something to do with the > IPC::SendSyncOption::UseFullySynchronousModeForTesting IPC send option that > is used by WK2 eventSender. When this flag is used for a Sync IPC A, every async IPC X that gets sent by the Sync IPC A reply handler becomes synchronous :S So in this case: Log(Pressing "z" with access key modi...) sends Async message to UIProcess eventSender.keyDown() sends Sync IPC to UIProcess with IPC::SendSyncOption::UseFullySynchronousModeForTesting As a result of the keydown, we do: Log(keydownevent dispatched) which should send an Async message to the UIProcess, but this becomes synchronous due to SendSyncOption::UseFullySynchronousModeForTesting. As a result, Log(keydownevent dispatched) may get processed by the UIProcess *before* Log(Pressing "z" with access key modi...) :( Created attachment 351007 [details]
Patch
Created attachment 351011 [details]
Patch
Created attachment 351015 [details]
Patch
Comment on attachment 351015 [details] Patch Attachment 351015 [details] did not pass mac-wk2-ews (mac-wk2): Output: https://webkit-queues.webkit.org/results/9374319 New failing tests: fast/css/pseudo-active-style-sharing-3.html fast/events/inactive-window-no-mouse-event.html loader/stateobjects/pushstate-size-iframe.html fullscreen/video-controls-timeline.html fast/css/pseudo-active-style-sharing-6.html fast/forms/file/input-file-write-files-using-open-panel.html fast/events/frame-detached-in-mousedown.html fast/css/pseudo-active-style-sharing-1.html fullscreen/full-screen-keyboard-enabled.html media/modern-media-controls/slider/slider-value.html fast/events/keydown-remove-frame.html fast/forms/placeholder-crash-with-scrollbar-corner.html loader/stateobjects/pushstate-size.html fast/visual-viewport/client-coordinates-relative-to-layout-viewport.html fast/events/can-click-element-on-page-with-active-pseudo-class-and-search-field.html fast/css/pseudo-active-style-sharing-5.html media/modern-media-controls/button/button-click-on-edges.html fast/css/pseudo-active-style-sharing-2.html Created attachment 351030 [details]
Archive of layout-test-results from ews106 for mac-sierra-wk2
The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews.
Bot: ews106 Port: mac-sierra-wk2 Platform: Mac OS X 10.12.6
Created attachment 351079 [details]
Patch
Comment on attachment 351079 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=351079&action=review > Source/WebKit/ChangeLog:7 > + Add a comment about what we're adding here too? Committed r236631: <https://trac.webkit.org/changeset/236631> This caused various tests to not complete: fast/css/pseudo-active-style-sharing-1.html fast/css/pseudo-active-style-sharing-2.html fast/css/pseudo-active-style-sharing-3.html fast/css/pseudo-active-style-sharing-4.html fast/css/pseudo-active-style-sharing-5.html fast/css/pseudo-active-style-sharing-6.html fast/events/can-click-element-on-page-with-active-pseudo-class-and-search-field.html fast/events/frame-detached-in-mousedown.html fast/events/inactive-window-no-mouse-event.html fast/events/keydown-remove-frame.html fast/forms/placeholder-crash-with-scrollbar-corner.html fast/visual-viewport/client-coordinates-relative-to-layout-viewport.html fullscreen/full-screen-keyboard-enabled.html fullscreen/video-controls-timeline.html loader/stateobjects/pushstate-size-iframe.html loader/stateobjects/pushstate-size.html media/modern-media-controls/button/button-click-on-edges.html media/modern-media-controls/slider/slider-value.html plugins/snapshotting/plugin-receives-click-event.html I'll roll it out. For example: https://build.webkit.org/results/Apple%20High%20Sierra%20Debug%20WK2%20(Tests)/r236643%20(5030)/results.html Re-opened since this is blocked by bug 190112 (In reply to Simon Fraser (smfr) from comment #13) > This caused various tests to not complete: > > fast/css/pseudo-active-style-sharing-1.html > fast/css/pseudo-active-style-sharing-2.html > fast/css/pseudo-active-style-sharing-3.html > fast/css/pseudo-active-style-sharing-4.html > fast/css/pseudo-active-style-sharing-5.html > fast/css/pseudo-active-style-sharing-6.html > fast/events/can-click-element-on-page-with-active-pseudo-class-and-search- > field.html > fast/events/frame-detached-in-mousedown.html > fast/events/inactive-window-no-mouse-event.html > fast/events/keydown-remove-frame.html > fast/forms/placeholder-crash-with-scrollbar-corner.html > fast/visual-viewport/client-coordinates-relative-to-layout-viewport.html > fullscreen/full-screen-keyboard-enabled.html > fullscreen/video-controls-timeline.html > loader/stateobjects/pushstate-size-iframe.html > loader/stateobjects/pushstate-size.html > media/modern-media-controls/button/button-click-on-edges.html > media/modern-media-controls/slider/slider-value.html > plugins/snapshotting/plugin-receives-click-event.html > > I'll roll it out. Weird. Ews was green. Thanks for rolling out, I will investigate on Monday. (In reply to Simon Fraser (smfr) from comment #13) > This caused various tests to not complete: > > fast/css/pseudo-active-style-sharing-1.html > fast/css/pseudo-active-style-sharing-2.html > fast/css/pseudo-active-style-sharing-3.html > fast/css/pseudo-active-style-sharing-4.html > fast/css/pseudo-active-style-sharing-5.html > fast/css/pseudo-active-style-sharing-6.html > fast/events/can-click-element-on-page-with-active-pseudo-class-and-search- > field.html > fast/events/frame-detached-in-mousedown.html > fast/events/inactive-window-no-mouse-event.html > fast/events/keydown-remove-frame.html > fast/forms/placeholder-crash-with-scrollbar-corner.html > fast/visual-viewport/client-coordinates-relative-to-layout-viewport.html > fullscreen/full-screen-keyboard-enabled.html > fullscreen/video-controls-timeline.html > loader/stateobjects/pushstate-size-iframe.html > loader/stateobjects/pushstate-size.html > media/modern-media-controls/button/button-click-on-edges.html > media/modern-media-controls/slider/slider-value.html > plugins/snapshotting/plugin-receives-click-event.html > > I'll roll it out. From what I see, it merely complained that those tests have no expected results, not that they were timing out or failing per say. (In reply to Chris Dumez from comment #17) > (In reply to Simon Fraser (smfr) from comment #13) > > This caused various tests to not complete: > > > > fast/css/pseudo-active-style-sharing-1.html > > fast/css/pseudo-active-style-sharing-2.html > > fast/css/pseudo-active-style-sharing-3.html > > fast/css/pseudo-active-style-sharing-4.html > > fast/css/pseudo-active-style-sharing-5.html > > fast/css/pseudo-active-style-sharing-6.html > > fast/events/can-click-element-on-page-with-active-pseudo-class-and-search- > > field.html > > fast/events/frame-detached-in-mousedown.html > > fast/events/inactive-window-no-mouse-event.html > > fast/events/keydown-remove-frame.html > > fast/forms/placeholder-crash-with-scrollbar-corner.html > > fast/visual-viewport/client-coordinates-relative-to-layout-viewport.html > > fullscreen/full-screen-keyboard-enabled.html > > fullscreen/video-controls-timeline.html > > loader/stateobjects/pushstate-size-iframe.html > > loader/stateobjects/pushstate-size.html > > media/modern-media-controls/button/button-click-on-edges.html > > media/modern-media-controls/slider/slider-value.html > > plugins/snapshotting/plugin-receives-click-event.html > > > > I'll roll it out. > > From what I see, it merely complained that those tests have no expected > results, not that they were timing out or failing per say. Ok yes, there is something weird going on as these tests do seem to have expected tests results: LayoutTests/fast/css/pseudo-active-style-sharing-1-expected.txt Created attachment 351250 [details]
Patch
The issue was that the "NotifyDone" IPC would still become synchronous when the IPC connection was in synchronous mode, and therefore could be processed before the OutputText asynchronous IPC. The tests would sometimes complete with no output as a result. Comment on attachment 351250 [details] Patch Rejecting attachment 351250 [details] from commit-queue. Failed to run "['/Volumes/Data/EWS/WebKit/Tools/Scripts/webkit-patch', '--status-host=webkit-queues.webkit.org', '--bot-id=webkit-cq-03', 'land-attachment', '--force-clean', '--non-interactive', '--parent-command=commit-queue', 351250, '--port=mac']" exit_code: 2 cwd: /Volumes/Data/EWS/WebKit Logging in as commit-queue@webkit.org... Fetching: https://bugs.webkit.org/attachment.cgi?id=351250&action=edit Fetching: https://bugs.webkit.org/show_bug.cgi?id=190052&ctype=xml&excludefield=attachmentdata Processing 1 patch from 1 bug. Updating working directory Processing patch 351250 from bug 190052. Fetching: https://bugs.webkit.org/attachment.cgi?id=351250 Failed to run "['git', 'svn', 'dcommit', '--rmdir']" exit_code: 1 cwd: /Volumes/Data/EWS/WebKit Committing to http://svn.webkit.org/repository/webkit/trunk ... M Source/WebKit/ChangeLog ERROR from SVN: Item is out of date: File '/trunk/Source/WebKit/ChangeLog' is out of date W: 58feecd202116e40367278a1bfa436ff49358423 and refs/remotes/origin/master differ, using rebase: :040000 040000 0fb40b159be9ec044e83f9b0a19815371ea0f068 f1e39de4dfb3a84f2125f3470accd9d338990f03 M Source :040000 040000 5f7163d142cbe02ea12ec3b913fe56304aec3c35 d78aa9b6cf6da2a632cd62cbfe6837473607de99 M Tools Current branch master is up to date. ERROR: Not all changes have been committed into SVN, however the committed ones (if any) seem to be successfully integrated into the working tree. Please see the above messages for details. Failed to run "['git', 'svn', 'dcommit', '--rmdir']" exit_code: 1 cwd: /Volumes/Data/EWS/WebKit Committing to http://svn.webkit.org/repository/webkit/trunk ... M Source/WebKit/ChangeLog ERROR from SVN: Item is out of date: File '/trunk/Source/WebKit/ChangeLog' is out of date W: 58feecd202116e40367278a1bfa436ff49358423 and refs/remotes/origin/master differ, using rebase: :040000 040000 0fb40b159be9ec044e83f9b0a19815371ea0f068 f1e39de4dfb3a84f2125f3470accd9d338990f03 M Source :040000 040000 5f7163d142cbe02ea12ec3b913fe56304aec3c35 d78aa9b6cf6da2a632cd62cbfe6837473607de99 M Tools Current branch master is up to date. ERROR: Not all changes have been committed into SVN, however the committed ones (if any) seem to be successfully integrated into the working tree. Please see the above messages for details. Failed to run "['git', 'svn', 'dcommit', '--rmdir']" exit_code: 1 cwd: /Volumes/Data/EWS/WebKit Updating OpenSource Current branch master is up to date. Full output: https://webkit-queues.webkit.org/results/9411910 Comment on attachment 351250 [details] Patch Clearing flags on attachment: 351250 Committed r236680: <https://trac.webkit.org/changeset/236680> All reviewed patches have been landed. Closing bug. |