Fingerprinting 2.0: Plugins #9435

pes10k · 2020-04-23T23:35:08Z

This is a sub-issue of the larger fingerprinting defense reorganization #8787

Plugins interfaces

NavigatorPlugins.plugins
navigator.plugins

Currently Brave reports two plugins, "Chrome PDF Plugin" and "Chrome PDF Viewer"

Word replacements:

Chrome: ["Chrome", "Chromium", "Brave", "Web", "Browser", "" (empty string), "OpenSource", "Online", "JavaScript"]
PDF: ["PDF", "Portable Document Format", "portable-document-format", "document", "doc", "PDF and PS", "com.adobe.pdf"]
Plugin: ["Viewer", "Renderer", "Display", "Plugin", "plug-in", "plug in", "extension", "" (empty string)]
Viewer: ["Viewer", "Renderer", "Display", "Plugin", "plug-in", "plug in", "extension", "" (empty string)]

Random plugin
(see https://source.chromium.org/chromium/chromium/src/+/master:third_party/blink/renderer/modules/plugins/dom_plugin.cc)

description: any 3 random words, possibly taken from chrome/browser/spellchecker/spellcheck_hunspell_dictionary.h
filename: 32 random characters, from [a-z]
name: any 3 random words, possibly taken from chrome/browser/spellchecker/spellcheck_hunspell_dictionary.h
length: 0 (needed for web compat reasons)
namedItem(): always null (again needed for web compat)

default protection:

Randomize both descriptions of plugins, under eTLD+1 seed, using random selections from the above word replacements
Randomize both filename to be 32 random characters, under eTLD+1 seed, from [a-z]
Include two randomly generated plugins, using the above schema
Randomize return order

max protection:

Include two randomly generated plugins, using the above schema
Randomize return order
Include no other plugin information

The text was updated successfully, but these errors were encountered:

Madis0 · 2020-05-22T18:33:20Z

I don't see how the randomization of names could help.

The issue is bigger than Brave - if the UA, rendering engine and other stuff already tell the world it's Chromium, plugin randomization would only make the user more distinct, because it is unlike any other Chromium.
So keeping it 1:1 with latest Chrome (which I think also has Native Client) would be the best bet.

Another option could be emulating Firefox/Tor browser by simply keeping it an empty array, but again, that needs more precautions overall.

pes10k · 2020-05-22T18:35:22Z

https://brave.com/whats-brave-done-for-my-privacy-lately-episode3/ gives some background why randomization is superior to fixed response, when you can do it w/o breaking things

pes10k · 2020-07-16T22:29:06Z

For QA, I've added a plugins test to https://dev-pages.brave.software/farbling.html

LaurenWags · 2020-08-17T19:18:52Z

Verified passed with

Brave | 1.13.73 Chromium: 84.0.4147.125 (Official Build) dev (64-bit)
-- | --
Revision | d0784639447f2e10d32ebaf9861092b20cfde286-refs/branch-heads/4147@{#1059}
OS | macOS Version 10.14.6 (Build 18G3020)

Verified using https://dev-pages.bravesoftware.com/farbling.html and https://dev-pages.brave.software/farbling.html

Confirmed with shields up, FP values are the same for "Plugins" per page, but the values are different between the two pages above

Note - logged #11278 as a follow-up for the Local Frame value being different from This Page and Remote Frame value

Default Shields (FP = Standard):

Modified Shields (FP = Strict):

Confirmed with shields up, if you close and relaunch the browser, and visit the pages again the values for "Plugins" are different (they are not retained between sessions)

See note about logged issue in above

Default Shields (FP = Standard):

Modified Shields (FP = Strict):

Confirmed with shields down, FP values for "Plugins" are the same for each individual item on a given page and are the same for both pages

Shields Down:

Confirmed with shields down, if you close and relaunch the browser, and visit the pages again, the values for "Plugins: are the same (they are retained between sessions)

Shields Down:

Verification passed on


Brave | 1.13.77 Chromium: 85.0.4183.69 (Official Build) dev (64-bit)
-- | --
Revision | 4554ea1a1171bd8d06951a4b7d9336afe6c59967-refs/branch-heads/4183@{#1426}
OS | Windows 10 OS Version 1903 (Build 18362.1016)

Verified using https://dev-pages.bravesoftware.com/farbling.html and https://dev-pages.brave.software/farbling.html
Confirmed with shields up (FP = Standard), FP values are the same for "Plugins" per page, but the values are different between the two pages above

https://dev-pages.bravesoftware.com/farbling.html	https://dev-pages.brave.software/farbling.html

Encountered Plugins FP blocking not working correctly when FP set to Standard #11278

(FP = Strict):

https://dev-pages.bravesoftware.com/farbling.html	https://dev-pages.brave.software/farbling.html

Confirmed with shields up (FP = Standard), if you close and relaunch the browser, and visit the pages again the values for "Plugins" are different (they are not retained between sessions)

https://dev-pages.bravesoftware.com/farbling.html	https://dev-pages.brave.software/farbling.html

(FP = Strict):

https://dev-pages.bravesoftware.com/farbling.html	https://dev-pages.brave.software/farbling.html

Confirmed with shields down, FP values for "Plugins" are the same for each individual item on a given page and are the same for both pages

https://dev-pages.bravesoftware.com/farbling.html	https://dev-pages.brave.software/farbling.html

Confirmed with shields down, if you close and relaunch the browser, and visit the pages again, the values for "Plugins: are the same (they are retained between sessions)

https://dev-pages.bravesoftware.com/farbling.html	https://dev-pages.brave.software/farbling.html

Verification passed on

Brave	1.13.79 Chromium: 85.0.4183.69 (Official Build) dev (64-bit)
Revision	4554ea1a1171bd8d06951a4b7d9336afe6c59967-refs/branch-heads/4183@{#1426}
OS	Ubuntu 18.04 LTS

Verified using https://dev-pages.bravesoftware.com/farbling.html and https://dev-pages.brave.software/farbling.html

Confirmed with shields up, FP values are the same for "Plugins" per page, but the values are different between the two pages above

Encountered #11278

Default Shields (FP = Standard):

Modified Shields (FP = Strict):

Confirmed with shields up, if you close and relaunch the browser, and visit the pages again the values for "Plugins" are different (they are not retained between sessions)

See note about logged issue in above

Default Shields (FP = Standard):

Modified Shields (FP = Strict):

Confirmed with shields down, FP values for "Plugins" are the same for each individual item on a given page and are the same for both pages

Shields Down:

Confirmed with shields down, if you close and relaunch the browser, and visit the pages again, the values for "Plugins: are the same (they are retained between sessions)

Shields Down:

srirambv · 2020-08-27T08:33:25Z

Verification passed on OnePlus 6T with Android 10 running 1.13.81 x64 build

Verified using https://dev-pages.bravesoftware.com/farbling.html and https://dev-pages.brave.software/farbling.html

Verified with shields up, FP values are same for "Plugins" on the same page but values are different on each page

https://dev-pages.brave.software/farbling.html	https://dev-pages.bravesoftware.com/farbling.html

Verified with shields up closing and relaunching the browser values for "Plugins" are not retained between sessions

https://dev-pages.brave.software/farbling.html	https://dev-pages.bravesoftware.com/farbling.html

Verified with shields down, FP values for "Plugins" are set to `00000000` as plugin list is empty

https://dev-pages.brave.software/farbling.html	https://dev-pages.bravesoftware.com/farbling.html

Logged https://github.com/[Android][Follow up #9435] Fingerprinting 2.0: Plugins not listed when shields is disabled #11460 for investigating no plugins when shields is disabled

Verified with shields down, FP values for "Plugins" value of "00000000" is retained between sessions

https://dev-pages.brave.software/farbling.html	https://dev-pages.bravesoftware.com/farbling.html

Verification passed on Samsung Tab A with Android 10 running 1.13.81 x64 build

Verified using https://dev-pages.bravesoftware.com/farbling.html and https://dev-pages.brave.software/farbling.html

Verified with shields up, FP values are same for "Plugins" on the same page but values are different on each page

https://dev-pages.brave.software/farbling.html	https://dev-pages.bravesoftware.com/farbling.html

Verified with shields up closing and relaunching the browser values for "Plugins" are not retained between sessions

https://dev-pages.brave.software/farbling.html	https://dev-pages.bravesoftware.com/farbling.html

Verified with shields down, FP values for "Plugins" are set to `00000000` as plugin list is empty

https://dev-pages.brave.software/farbling.html	https://dev-pages.bravesoftware.com/farbling.html

Logged https://github.com/[Android][Follow up #9435] Fingerprinting 2.0: Plugins not listed when shields is disabled #11460 for investigating no plugins when shields is disabled

Verified with shields down, FP values for "Plugins" value of "00000000" is retained between sessions

https://dev-pages.brave.software/farbling.html	https://dev-pages.bravesoftware.com/farbling.html

Due to the removal of flash, navigator.plugins will always report an empty array from now on, so there's no need for Brave to implement a farbling mechanism for such array any more. Thus, we can now remove all the code added to implement that and drop a couple of patches to upstream's Chromium in the way too. Main related issues in Brave's GitHub: * Fingerprinting 2.0: Plugins brave/brave-browser#9435 * Fingerprinting 2.0: Plugins (word replacement) brave/brave-browser#10597 Chromium change: https://chromium.googlesource.com/chromium/src.git/+/f8fe422e0e8a026f73f9d74d26f75c3c93102030 commit f8fe422e0e8a026f73f9d74d26f75c3c93102030 Author: Mason Freed <masonfreed@chromium.org> Date: Wed Jan 20 18:50:56 2021 +0000 Empty out navigator.plugins and navigator.mimeTypes With this CL, navigator.plugins and navigator.mimeTypes will both return empty arrays. With the removal of Flash, there is no longer the need to return anything for navigator.plugins and navigator.mimeTypes. These APIs were used primarily for a) probing for Flash player support, or b) fingerprinting. As such, we'd like to return empty for these two properties. Gecko already switched to returning *only* Flash starting with Firefox 52, and returning empty as of Firefox 85. I2S: https://groups.google.com/a/chromium.org/g/blink-dev/c/bbxAGu90LgM Fixed: 1164635

Due to the removal of flash, navigator.plugins will always report an empty array from now on, so Brave needs to adapt its farbling mechanism to work like this depending on the selected behavior: * "off": return the real navigators.plugins, that is, an empty array. * "balanced" or "maximum": return 2 additional fake plugins. Additionally, we need to adapt the browser tests to account for this new scenario, and test that indeed the expectation of navigator.plugins returning an empty array stays true. Issues in Brave's GitHub related to the implementation of this feature: * Fingerprinting 2.0: Plugins brave/brave-browser#9435 * Fingerprinting 2.0: Plugins (word replacement) brave/brave-browser#10597 Chromium change: https://chromium.googlesource.com/chromium/src.git/+/f8fe422e0e8a026f73f9d74d26f75c3c93102030 commit f8fe422e0e8a026f73f9d74d26f75c3c93102030 Author: Mason Freed <masonfreed@chromium.org> Date: Wed Jan 20 18:50:56 2021 +0000 Empty out navigator.plugins and navigator.mimeTypes With this CL, navigator.plugins and navigator.mimeTypes will both return empty arrays. With the removal of Flash, there is no longer the need to return anything for navigator.plugins and navigator.mimeTypes. These APIs were used primarily for a) probing for Flash player support, or b) fingerprinting. As such, we'd like to return empty for these two properties. Gecko already switched to returning *only* Flash starting with Firefox 52, and returning empty as of Firefox 85. I2S: https://groups.google.com/a/chromium.org/g/blink-dev/c/bbxAGu90LgM Fixed: 1164635 GOOD

Due to the removal of flash, navigator.plugins will always report an empty array from now on, so Brave needs to adapt its farbling mechanism to work like this depending on the selected behavior: * "off": return the real navigators.plugins, that is, an empty array. * "balanced" or "maximum": return 2 randomly generated fake plugins. Additionally, we need to adapt the browser tests to account for this new scenario, and test that indeed the expectation of navigator.plugins returning an empty array stays true. Issues in Brave's GitHub related to the implementation of this feature: * Fingerprinting 2.0: Plugins brave/brave-browser#9435 * Fingerprinting 2.0: Plugins (word replacement) brave/brave-browser#10597 Chromium change: https://chromium.googlesource.com/chromium/src.git/+/f8fe422e0e8a026f73f9d74d26f75c3c93102030 commit f8fe422e0e8a026f73f9d74d26f75c3c93102030 Author: Mason Freed <masonfreed@chromium.org> Date: Wed Jan 20 18:50:56 2021 +0000 Empty out navigator.plugins and navigator.mimeTypes With this CL, navigator.plugins and navigator.mimeTypes will both return empty arrays. With the removal of Flash, there is no longer the need to return anything for navigator.plugins and navigator.mimeTypes. These APIs were used primarily for a) probing for Flash player support, or b) fingerprinting. As such, we'd like to return empty for these two properties. Gecko already switched to returning *only* Flash starting with Firefox 52, and returning empty as of Firefox 85. I2S: https://groups.google.com/a/chromium.org/g/blink-dev/c/bbxAGu90LgM Fixed: 1164635

pes10k added feature/shields/fingerprint The fingerprinting (aka: "device recognition") protection provided in Shields privacy-pod Feature work for the Privacy & Web Compatibility pod privacy labels Apr 24, 2020

pes10k mentioned this issue Apr 24, 2020

Fingerprinting Protections v2: Farbling and cross-origin #8787

Closed

10 tasks

pilgrim-brave mentioned this issue Jul 1, 2020

Implement navigator.plugins farbling brave/brave-core#5989

Merged

32 tasks

pilgrim-brave closed this as completed in brave/brave-core#5989 Jul 7, 2020

pes10k reopened this Jul 7, 2020

pes10k closed this as completed Jul 7, 2020

This was referenced Jul 7, 2020

Fingerprinting 2.0: Plugins (word replacement) #10597

Closed

Fingerprinting 2.0: Plugins (mimetype) #10718

Closed

pes10k added this to the 1.13.x - Nightly milestone Jul 16, 2020

pes10k added QA/Yes release-notes/include labels Jul 16, 2020

bbondy added OS/Android Fixes related to Android browser functionality OS/Desktop labels Aug 7, 2020

LaurenWags mentioned this issue Aug 17, 2020

Plugins FP blocking not working correctly when FP set to Standard #11278

Closed

LaurenWags added the QA Pass-macOS label Aug 17, 2020

GeetaSarvadnya added the QA Pass-Win64 label Aug 20, 2020

btlechowski added the QA Pass-Linux label Aug 21, 2020

LaurenWags mentioned this issue Aug 26, 2020

[Desktop] Release Notes for 1.13.x (Release Channel) #11454

Closed

srirambv mentioned this issue Aug 27, 2020

[Android][Follow up #9435] Fingerprinting 2.0: Plugins not listed when shields is disabled #11460

Closed

srirambv added QA Pass - Android ARM QA Pass - Android Tab labels Aug 27, 2020

This was referenced Sep 4, 2020

[Android] Release Notes for Android 1.13.x Release #11596

Closed

[Android] Issues to be verified from 1.13.x Release #1 #11636

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fingerprinting 2.0: Plugins #9435

Fingerprinting 2.0: Plugins #9435

pes10k commented Apr 23, 2020 •

edited

Loading

Madis0 commented May 22, 2020

pes10k commented May 22, 2020

pes10k commented Jul 16, 2020

LaurenWags commented Aug 17, 2020 •

edited by btlechowski

Loading

srirambv commented Aug 27, 2020

Fingerprinting 2.0: Plugins #9435

Fingerprinting 2.0: Plugins #9435

Comments

pes10k commented Apr 23, 2020 • edited Loading

Plugins interfaces

Madis0 commented May 22, 2020

pes10k commented May 22, 2020

pes10k commented Jul 16, 2020

LaurenWags commented Aug 17, 2020 • edited by btlechowski Loading

srirambv commented Aug 27, 2020

pes10k commented Apr 23, 2020 •

edited

Loading

LaurenWags commented Aug 17, 2020 •

edited by btlechowski

Loading