pointer tests behind user gestures

[Thorin-Oakenpants]

Notes

• fyi @patrickhlauke
• note: because this is behind a user gesture, it is not part of the section or global fingerprint (FP)
• initially I had a full screen opaque overlay (display none, but flipped to block after the FP finishes, and reflipped on reruns), but if dom.w3c_pointer_events.enabled (removed in FF87) is false, then no-one can select or click anything (e.g. Tor Browser), which seems problematic
  • so instead I made it the tiny bit you see above (it's actually the width of the table cell)
• this function is only called once, the listener is permanent

  • TZP/js/devices.js

    Line 411 in 97d6aea

    function get_pointer_event() {

  • another ugly code example from yours truly, I'm sure it could be more elegant

RFP patches

• 1363508 spoof/suppress Pointer Events (FF64)
• 1492766 spoof pointerEvent.pointerid (FF65)

ToDo

• test FF on android
  • DONE if android change test to tap here changed to mouse/tap
• add RFP notation based on isVer (FF64/FF65) - we should get consistent results on desktop vs android (and we already know the OS), IDK about touch pads, I think RFP treats them as Linux
• where is pointerid?
  • what other pointer tests with entropy can we add

Some results

height  mozInputSource  pressure     tiltX tiltY twist width
   isPrimary  pointerType  tangentialPressure 
       |             |            |
  1 | true | 1 |   mouse | 0   |  0   | 0 | 0 | 0 | 1  = my desktop regardless of RFP
  0 | true | 5 |   touch | 1   |  0   | 0 | 0 | 0 | 0  = my android without RFP
  1 | true | 5 |   mouse | 0.5 |  0   | 0 | 0 | 0 | 0  = my android with RFP

[Thorin-Oakenpants]

@tomrittervg ... pressure and width do not match RFP desktop vs RFP android .. edit and mozInputSource leaks