Issue metadata
Sign in to add a comment
|
PlzNavigate fails process bindings tests |
||||||||||||||||||||||||
Issue descriptionThe process bindings tests org.chromium.chrome.browser.BindingManagerIntegrationTest#testRestoreSharedRenderer org.chromium.chrome.browser.BindingManagerIntegrationTest#testCrashInForeground are failing on Chromium for Android with PlzNavigate. To repro, $ git cl patch 2385413004 // this forces PlzNavigate without requiring a switch $ ./out/Debug/bin/run_chrome_public_test_apk_incremental -f "*RestoreSharedRenderer*" // or "*CrashInForeground*"
,
Mar 11 2017
,
Mar 14 2017
Copying from offline conversation with John and Maria. Found a suspected place what could be the reason the following test fails org.chromium.chrome.browser.BindingManagerIntegrationTest#testCrashInForeground (line : 401) In this test, we load a URL in a tab, crash it intentionally and reload it again. In non-PlzNavigate mode, when a new renderer is starting after the crash, we call RenderWidgetHostView::Show. https://cs.chromium.org/chromium/src/content/browser/frame_host/render_frame_host_manager.cc?rcl=062df28033e325db34be74c68da1f2bec888229d&l=258 I don't see that called for PlzNavigate mode. Not sure what is expected here. As a result, we get to the following state in PlzNavigate mode with should_background = true and hence the new render process is not foregrounded. https://cs.chromium.org/chromium/src/content/browser/renderer_host/render_process_host_impl.cc?rcl=11e28fd6b9380b77273db51ef0b6ccc7ea341944&l=2831
,
Mar 17 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/8e07fc9bf21eaec35f94de57f9ba20cc5a9b009e commit 8e07fc9bf21eaec35f94de57f9ba20cc5a9b009e Author: jam <jam@chromium.org> Date: Fri Mar 17 00:34:14 2017 Ensure that with PlzNavigate, reloads after process crash display the RenderView. BUG= 699384 CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation Review-Url: https://codereview.chromium.org/2753093002 Cr-Commit-Position: refs/heads/master@{#457626} [modify] https://crrev.com/8e07fc9bf21eaec35f94de57f9ba20cc5a9b009e/chrome/browser/crash_recovery_browsertest.cc [modify] https://crrev.com/8e07fc9bf21eaec35f94de57f9ba20cc5a9b009e/content/browser/frame_host/render_frame_host_manager.cc [modify] https://crrev.com/8e07fc9bf21eaec35f94de57f9ba20cc5a9b009e/content/browser/frame_host/render_frame_host_manager.h
,
Mar 17 2017
I fixed the issue noted, but the BindingManagerIntegrationTest#testCrashInForeground still fails on android. Can you take a look?
,
Mar 18 2017
Thanks! I had to fix the tests little bit to wait till the loading is complete. After that these two tests seem to pass. I will upload a CL shortly. CL in progress: https://codereview.chromium.org/2759663003
,
Mar 18 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/a627a2107cb27b4707f1f3b4c741636d8b7c5ace commit a627a2107cb27b4707f1f3b4c741636d8b7c5ace Author: shaktisahu <shaktisahu@chromium.org> Date: Sat Mar 18 21:06:22 2017 Fixed process binding tests Changed tests to wait for tab getting loaded completely after a reload. BUG= 699384 Review-Url: https://codereview.chromium.org/2759663003 Cr-Commit-Position: refs/heads/master@{#457979} [modify] https://crrev.com/a627a2107cb27b4707f1f3b4c741636d8b7c5ace/chrome/android/javatests/src/org/chromium/chrome/browser/BindingManagerIntegrationTest.java
,
Mar 18 2017
,
Nov 7 2017
,
Nov 7 2017
Apologies, applied the wrong component in bulk. |
|||||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||||
Comment 1 by cblume@chromium.org
, Mar 8 2017