1617380 - 5% frame drops on HD YouTube videos with privacy.resistFingerprinting set to true

Status: RESOLVED INVALID

(Core :: Privacy: Anti-Tracking, defect, P3)

Description

[sworddragon2]

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:68.0) Gecko/20100101 Firefox/68.0

Steps to reproduce:

1. Set privacy.resistFingerprinting to true.
2. Play a YouTube video on HD settings (for example https://www.youtube.com/watch?v=FYCjJHhxONc on 1080p).
3. Right click the video and enable the YouTube video player statistics.

Actual results:

5% of the frames are being dropped.

Expected results:

The straight 5% looks too suspicious and it might be very intended for entropy elimination by obfuscating the clients playback performance. On assuming this is being intended:

• If the dropped frames are only reported (but not being dropped):
  • I guess this ticket can then be simply closed as invalid.
• If the frames are indeed being dropped:

  • I get the straight 5% frame drop independent if hardware acceleration is being used or not. Does a website actually know if a client uses hardware acceleration and thus can this information safely be revealed?

  • If so:

    • We probably can use then a lower frame drop rate for hardware accelerated clients.

  • If not:

    • Isn't a plain 5% for 720p and 1080p too high for todays systems? Even if I turn off hardware acceleration my CPU from 2012 (AMD Phenom II X6 1045T) got only 1/9058 frame drops in the above video. While in 2012 there were probably some more weak CPU's than mine casually around it also utilized only 33% on 1080p.

  • Alternatively eventually a stepping could be used (similar to the resolution stepping) to avoid too high frame drops on mid- and highend systems. Also a stepping would probably help obfuscating the performance for clients that have a higher frame drop than the current 5% and would allow us to also use it for non-HD videos.

Comment 1

[Alice0775 White]

I can reproduce the 5% frame drop issue on Nightly75.0a1 Windows10.

Comment 2

[Tom Ritter [:tjr] (OOTO until 4/30 at least)]

(In reply to sworddragon2 from comment #0)

• If the dropped frames are only reported (but not being dropped):
  • I guess this ticket can then be simply closed as invalid.

Good intuition :) Bug 1369309

Comment 3

[sworddragon2]

(In reply to Tom Ritter [:tjr] (ni for response to sec-[approval|rating|advisories|cve]) from comment #2)

Good intuition :) Bug 1369309

Thanks for the link - this also explains why different values for different resolutions are being reported. So all seems to be fine here.