VrShellNavigationTest#test2dTo2d seems broken in 3 completely unrelated ways.

First, it's failing 100% of the time on the Nougat bot. It first occurred in https://ci.chromium.org/buildbot/chromium.android/Nougat%20Phone%20Tester/3788, and the only VR-related CL there is a DOFF flow change. In this case, it's failing because it's failing to properly load the chrome://history page. The following is provided as the failure reason:

java.lang.AssertionError: Page did not load.  Tab information at time of failure -- url: file:///storage/emulated/0/chromium_tests_root/chrome/test/data/vr/e2e_test_files/html/test_navigation_2d_page2.html, final URL: chrome://history/, load progress: 100, is loading: false, web contents loading: false

Looking at the logcat output, the following lines just before the failure seem slightly suspicious, although I don't know how this would relate to page loading:

02-22 04:20:52.435 11158 27753 I VrCtl.StateMachine: Restarting for controller GoogleDaydreamControllerV1
02-22 04:20:52.435 11158 27753 I VrCtl.StateMachine: Preparing controller GoogleDaydreamControllerV1
02-22 04:20:52.436 11158 27753 I VrCtl.StateMachine: Controller GoogleDaydreamControllerV1 state change STATE_ERROR --> STATE_PREPARING
02-22 04:20:52.436 11158 27753 I VrCtl.ControllerSelector: Creating ControllerDevice with driver: DRIVER_P6
02-22 04:20:52.436 11158 27753 I VrCtl.ControllerDriverFactory: Creating ControllerDevice for driver: DRIVER_P6
02-22 04:20:52.437 11158 27753 I VrCtl.PaprikaController: Soft disconnect mode: 
02-22 04:20:52.447 11158 27753 E VrCtl.ControllerSelector: Controller unexpectedly absent from bonded device list: [name=DRIVER_P6,isEmulator=false,address=,usesBluetooth=false]
02-22 04:20:52.447 11158 27753 E VrCtl.StateMachine: Controller GoogleDaydreamControllerV1 experienced error: ERROR_CONTROLLER_NOT_BONDED
02-22 04:20:52.448 11158 27753 I VrCtl.StateMachine: Controller GoogleDaydreamControllerV1 state change STATE_PREPARING --> STATE_ERROR
02-22 04:20:52.449 11158 27753 W VrCtl.ControllerService: No controller found since the restart.
02-22 04:20:52.449 11158 27753 I VrCtl.ListenersManager: Controller GoogleDaydreamControllerV1 state updated to: SCANNING
02-22 04:20:52.449 11158 11158 I VrCtl.ControllerService: Not launching recovery flow for controller: GoogleDaydreamControllerV1 missingTimeMillis: 0
02-22 04:20:52.452 11158 27753 I VrCtl.ListenersManager: Controller GoogleDaydreamControllerV1 state updated to: DISCONNECTED
02-22 04:20:57.454 11158 27753 I VrCtl.StateMachine: Restarting for controller GoogleDaydreamControllerV1
02-22 04:20:57.454 11158 27753 I VrCtl.StateMachine: Preparing controller GoogleDaydreamControllerV1
02-22 04:20:57.454 11158 27753 I VrCtl.StateMachine: Controller GoogleDaydreamControllerV1 state change STATE_ERROR --> STATE_PREPARING
02-22 04:20:57.454 11158 27753 I VrCtl.ControllerSelector: Creating ControllerDevice with driver: DRIVER_P6
02-22 04:20:57.455 11158 27753 I VrCtl.ControllerDriverFactory: Creating ControllerDevice for driver: DRIVER_P6
02-22 04:20:57.456 11158 27753 I VrCtl.PaprikaController: Soft disconnect mode: 
02-22 04:20:57.464 11158 27753 E VrCtl.ControllerSelector: Controller unexpectedly absent from bonded device list: [name=DRIVER_P6,isEmulator=false,address=,usesBluetooth=false]
02-22 04:20:57.464 11158 27753 E VrCtl.StateMachine: Controller GoogleDaydreamControllerV1 experienced error: ERROR_CONTROLLER_NOT_BONDED
02-22 04:20:57.465 11158 27753 I VrCtl.StateMachine: Controller GoogleDaydreamControllerV1 state change STATE_PREPARING --> STATE_ERROR
02-22 04:20:57.466 11158 27753 W VrCtl.ControllerService: No controller found since the restart.
02-22 04:20:57.466 11158 27753 I VrCtl.ListenersManager: Controller GoogleDaydreamControllerV1 state updated to: SCANNING
02-22 04:20:57.467 11158 11158 I VrCtl.ControllerService: Not launching recovery flow for controller: GoogleDaydreamControllerV1 missingTimeMillis: 0
02-22 04:20:57.467 11158 27753 I VrCtl.ListenersManager: Controller GoogleDaydreamControllerV1 state updated to: DISCONNECTED

The post-failure screenshot shows that something's clearly wrong, as the content quad is completely black https://storage.cloud.google.com/chromium-result-details/f20440473962d8e4f619f5b7e5d6e3292fba01f1

Trying to reproduce this issue locally on O resulted in the two other failures.

If I run the test while the phone is in landscape mode, the browser crashes due to hitting the OnMainThread() DCHECK in VrGLThread::ShowExitVrPrompt at https://cs.chromium.org/chromium/src/chrome/browser/android/vr_shell/vr_gl_thread.cc?q=vr_gl_thread.cc&sq=package:chromium&dr&l=363. This is weird because we shouldn't need to show the exit prompt during the test.

If I run the test while the phone is in portrait mode, the history page is clearly messed up (super zoomed in), and the test fails due to:

java.lang.AssertionError: Two navigations showed up in history expected:<2> but was:<1>

This sort of makes sense, as only one history item is actually being shown (the way we check the number of history items might be affected by what's being actually displayed), but the history page weirdness that causes that is unexpected.