The layout tests for sensors are already using shared buffer, and the JS mocks for the backends (pls see https://cs.chromium.org/chromium/src/third_party/WebKit/LayoutTests/sensor/resources/sensor-helpers.js) Is there anything missing?

hmm... I think this issue is for updating some test code related to device motion and orientation event, not the generic sensor.

I agree, shared buffer support appears to be ready. Repurposing this bug to track rewriting LayoutTests/device_orientation/resources/device-orientation-helpers.js to use the Generic Sensor mocking infrastructure so we can remove the testRunner methods for setting fake sensor data.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/fb44d3a0e6db2c18c4cb6d5f2fb1550ee9b8e268

commit fb44d3a0e6db2c18c4cb6d5f2fb1550ee9b8e268
Author: Robbie McElrath <rmcelrath@chromium.org>
Date: Fri May 25 18:48:50 2018

Make sensor event pump classes use mojom service in layout tests.

This is needed to migrate sensor layout tests to using mock mojom services
instead of testRunner.setMockDevice[Motion|Orientation]. Currently the event
pump classes don't bind to a mojom service if they're running as part of a
layout test, but they'll need to in order to use a mock mojom service to provide
the testing data.

This will make existing layout tests try to connect to the generic sensor
provider mojom service, which could break the existing tests if the null-event
is fired when the service isn't found. This isn't actually an issue with the
existing layout tests because of the way the sensor event pump classes handle
missing mojom services. DeviceSensorEventPump::HandleSensorProviderError is
called when the SensorProvider service can't be found, which resets the mojom
pipe without calling OnSensorCreated, which would lead to the null-event being
fired.

Eventually we should fix the sensor event pump classes to send the null-event
when the sensor provider mojom service isn't found, but for now the fact that we
don't allows us to migrate the sensor layout tests piecemeal, which is good.

There are two tests that this affects. One of them is a wpt test that we
currently timeout on, but now fails. I'll fix that in a followup CL. The other
test this breaks is testing behavior that goes against the spec, so I updated
that test to actually test what it says it's testing, and to match the spec.

Bug:  774183 ,  802815 
Change-Id: I376d94a7e7f35aefdb4b586050a03e198988e8be
Reviewed-on: https://chromium-review.googlesource.com/1060380
Reviewed-by: Tim Volodine <timvolodine@chromium.org>
Reviewed-by: Jun Cai <juncai@chromium.org>
Reviewed-by: Reilly Grant <reillyg@chromium.org>
Commit-Queue: Robbie McElrath <rmcelrath@chromium.org>
Cr-Commit-Position: refs/heads/master@{#561959}
[modify] https://crrev.com/fb44d3a0e6db2c18c4cb6d5f2fb1550ee9b8e268/content/renderer/device_sensors/device_motion_event_pump.cc
[modify] https://crrev.com/fb44d3a0e6db2c18c4cb6d5f2fb1550ee9b8e268/content/renderer/device_sensors/device_motion_event_pump.h
[modify] https://crrev.com/fb44d3a0e6db2c18c4cb6d5f2fb1550ee9b8e268/content/renderer/device_sensors/device_motion_event_pump_unittest.cc
[modify] https://crrev.com/fb44d3a0e6db2c18c4cb6d5f2fb1550ee9b8e268/content/renderer/device_sensors/device_orientation_event_pump.cc
[modify] https://crrev.com/fb44d3a0e6db2c18c4cb6d5f2fb1550ee9b8e268/content/renderer/device_sensors/device_orientation_event_pump.h
[modify] https://crrev.com/fb44d3a0e6db2c18c4cb6d5f2fb1550ee9b8e268/content/renderer/device_sensors/device_orientation_event_pump_unittest.cc
[modify] https://crrev.com/fb44d3a0e6db2c18c4cb6d5f2fb1550ee9b8e268/third_party/WebKit/LayoutTests/TestExpectations
[modify] https://crrev.com/fb44d3a0e6db2c18c4cb6d5f2fb1550ee9b8e268/third_party/WebKit/LayoutTests/device_orientation/motion/add-during-dispatch.html

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/8bc33b3f35e74d698d49bb31934f6d409b7cf84d

commit 8bc33b3f35e74d698d49bb31934f6d409b7cf84d
Author: Robbie McElrath <rmcelrath@chromium.org>
Date: Fri May 25 19:58:09 2018

Make deviceorientationabsoluteevent WPT test match spec

The deviceorientationabsoluteevent WPT test currently asserts that the
'absolute' property of the deviceorientationabsolute event is false.

The spec states that:
https://w3c.github.io/deviceorientation/spec-source-orientation.html#deviceorientationabsolute
"The deviceorientationabsolute event is completely analogous to the
deviceorientation event, except additional sensors like the magnetometer can be
used to provide an absolute orientation. The absolute property must be set to
true."

This updates the test to assert that absolute is set to true.

Bug:  774183 ,  802815 
Change-Id: I3694e2b774aa0e4a8a9f7d0328e2aad8b37775e6
Reviewed-on: https://chromium-review.googlesource.com/1062650
Commit-Queue: Robbie McElrath <rmcelrath@chromium.org>
Reviewed-by: Jun Cai <juncai@chromium.org>
Reviewed-by: Reilly Grant <reillyg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#561984}
[modify] https://crrev.com/8bc33b3f35e74d698d49bb31934f6d409b7cf84d/third_party/WebKit/LayoutTests/TestExpectations
[modify] https://crrev.com/8bc33b3f35e74d698d49bb31934f6d409b7cf84d/third_party/WebKit/LayoutTests/external/wpt/orientation-event/deviceorientationabsoluteevent.html