http/tests/inspector/target/target-events-for-provisional-page.html added in https://trac.webkit.org/changeset/251494/webkit is a flakey failure when running with other tests. Probable cause: Some side effect of previous tests that disable PSON. Flakiness Dashboard: https://webkit-test-results.webkit.org/dashboards/flakiness_dashboard.html#showAllRuns=true&tests=http%2Ftests%2Finspector%2Ftarget%2Ftarget-events-for-provisional-page.html
The test reliably fails when running after http/tests/dom/new-window-can-target-opener.html. Tools/Scripts/run-webkit-tests --no-show-results --no-retry-failures --child-processes=1 http/tests/dom/new-window-can-target-opener.html http/tests/inspector/target/target-events-for-provisional-page.html
<rdar://problem/56988642>
The problem is that new-window-can-target-opener-win.html opened in a child window calls open() on its opener window, i.e. the test page itself (new-window-can-target-opener.html). It in turn triggers Page::setOpenedByDOMWithOpener() on the parent page. The next test will reuse the same instance of PlatformWebView and consequently the same Page in the WebProcess. Since instance of the page is the same it still has Page::openedByDOMWithOpener() == true. Because of that[1] cross-origin navigation in the next test page will reuse same process. I don't know if this is intentional behavior. I'm going to add <!-- webkit-test-runner [ enableProcessSwapOnWindowOpen=true ] --> as a temporary workaround for the flaky inspector test but we should probably never reuse a page which was openedByDOM. [1] https://trac.webkit.org/browser/webkit/trunk/Source/WebKit/UIProcess/WebProcessPool.cpp?rev=252014#L2139 [2] https://trac.webkit.org/browser/webkit/trunk/LayoutTests/http/tests/dom/resources/new-window-can-target-opener-win.html#L19
Created attachment 383075 [details] Patch
Comment on attachment 383075 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=383075&action=review rs=me > LayoutTests/ChangeLog:11 > + that may run before (the test may reuse Page instance from the previous test on which > + Page::openedByDOMWithOpener==true). This seems like a bug. Can you file a separate bug to fix this please?
Filed https://bugs.webkit.org/show_bug.cgi?id=203981
Comment on attachment 383075 [details] Patch Clearing flags on attachment: 383075 Committed r252214: <https://trac.webkit.org/changeset/252214>
All reviewed patches have been landed. Closing bug.