Skip to content
Snippets Groups Projects
New look: Off

Make sure WebExtensions are spared from JIT disabling in higher security settings (Medium-High)

    • Closed (moved) Issue created by cypherpunks @cypherpunks

    This could for example negatively affect HTTPS Everywhere's performance. I have however no data on whether JIT is disabled for WebExtensions in this case.

    To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information

    Child items ...

    • Activity

    • cypherpunks added TorBrowserTeam202006 actualpoints::0.5 component::applications/tor browser owner::acat points::1 priority::medium resolution::wontfix severity::major status::closed tbb-9.5a7-want tbb-performance type::defect labels

      added TorBrowserTeam202006 actualpoints::0.5 component::applications/tor browser owner::acat points::1 priority::medium resolution::wontfix severity::major status::closed tbb-9.5a7-want tbb-performance type::defect labels

    • Trac
      Trac @tracbot ·

      Trac:
      Username: tokotoko
      Cc: N/A to fdsfgs@krutt.org

    • C
      cypherpunks @cypherpunks ·
      Author

      Trac:
      Keywords: N/A deleted, ff59-esr added

    • C
      cypherpunks @cypherpunks ·
      Author

      I just confirmed it, at Medium-High security setting, JIT is disabled for WebExtensions resulting in lower performance.

      Trac:
      Severity: Normal to Major
      Keywords: ff59-esr deleted, tbb-performance added

    • C
      cypherpunks @cypherpunks ·
      Author

      Can this be given a higher prio? It's seems to severely affect my browsing as some tabs take a while to even start loading with Safer-Safest (seems related to NoScript doing its work).

    • Georg Koppen
      Georg Koppen @gk ·

      Replying to cypherpunks:

      Can this be given a higher prio? It's seems to severely affect my browsing as some tabs take a while to even start loading with Safer-Safest (seems related to NoScript doing its work).

      Is that reproducible (e.g. by loading a particular URL)? I am curious where the NoScript relation is coming from.

      Trac:
      Status: new to needs_information

    • C
      cypherpunks @cypherpunks ·
      Author

      Replying to gk:

      Is that reproducible (e.g. by loading a particular URL)? I am curious where the NoScript relation is coming from. Thanks for taking interest in this. It's a bit difficult to reproduce immediately but sometimes when I'm on a page and I ctrl+click on a link so that it opens in a new tab, the cpu usage gets to 100% on one core for about 10sec and it's taking 20sec or so just to starting loading up the page, from the browser console I see NoScript doing its thing (i.e. the CSP hack) during that time. (on 8.0a9 of course) It doesn't happen always so it's difficult to give clear ways to reproduce other than "just browse with safer security setting and ctrl+click on links".

    • Georg Koppen
      Georg Koppen @gk ·

      Replying to cypherpunks:

      Replying to gk:

      Is that reproducible (e.g. by loading a particular URL)? I am curious where the NoScript relation is coming from. Thanks for taking interest in this. It's a bit difficult to reproduce immediately but sometimes when I'm on a page and I ctrl+click on a link so that it opens in a new tab, the cpu usage gets to 100% on one core for about 10sec and it's taking 20sec or so just to starting loading up the page, from the browser console I see NoScript doing its thing (i.e. the CSP hack) during that time. (on 8.0a9 of course) It doesn't happen always so it's difficult to give clear ways to reproduce other than "just browse with safer security setting and ctrl+click on links".

      Is that on Windows? Or Linux? Or...

    • C
      cypherpunks @cypherpunks ·
      Author

      Replying to gk:

      Is that on Windows? Or Linux? Or... Yeah Linux.

    • C
      cypherpunks @cypherpunks ·
      Author

      Maybe it has nothing to do with JIT after all (though that doesn't mean that this ticket shouldn't be addressed) but such relief came when ye finally catch the fish:

      [NoScript] Cannot collect noscript activity data Error: Could not establish connection. Receiving end does not exist.
Stack trace:
collectSeen@moz-extension://[NoScript's UUID]/bg/main.js:265:38
 Could not establish connection. Receiving end does not exist. collectSeen@moz-extension://[NoScript's UUID]/bg/main.js:265:38
  log.js:12:5
	error moz-extension://[NoScript's UUID]/lib/log.js:12:5
	collectSeen moz-extension://[NoScript's UUID]/bg/main.js:274:10
[NoScript] Cannot collect noscript activity data Error: Could not establish connection. Receiving end does not exist.
Stack trace:
collectSeen@moz-extension://[NoScript's UUID]/bg/main.js:265:38
 Could not establish connection. Receiving end does not exist. collectSeen@moz-extension://[NoScript's UUID]/bg/main.js:265:38
  log.js:12:5
	error moz-extension://[NoScript's UUID]/lib/log.js:12:5
	collectSeen moz-extension://[NoScript's UUID]/bg/main.js:274:10
[NoScript] Cannot collect noscript activity data Error: Could not establish connection. Receiving end does not exist.
Stack trace:
collectSeen@moz-extension://[NoScript's UUID]/bg/main.js:265:38
 Could not establish connection. Receiving end does not exist. collectSeen@moz-extension://[NoScript's UUID]/bg/main.js:265:38
  log.js:12:5
	error moz-extension://[NoScript's UUID]/lib/log.js:12:5
	collectSeen moz-extension://[NoScript's UUID]/bg/main.js:274:10
[NoScript] Cannot collect noscript activity data Error: Could not establish connection. Receiving end does not exist.
Stack trace:
collectSeen@moz-extension://[NoScript's UUID]/bg/main.js:265:38
 Could not establish connection. Receiving end does not exist. collectSeen@moz-extension://[NoScript's UUID]/bg/main.js:265:38
  log.js:12:5
	error moz-extension://[NoScript's UUID]/lib/log.js:12:5
	collectSeen moz-extension://[NoScript's UUID]/bg/main.js:274:10
NS_ERROR_NOT_AVAILABLE: Component returned failure code: 0x80040111 (NS_ERROR_NOT_AVAILABLE) [nsIWebNavigation.loadURIWithOptions]  browser-child.js:359
[NoScript] Cannot collect noscript activity data Error: Could not establish connection. Receiving end does not exist.
Stack trace:
collectSeen@moz-extension://[NoScript's UUID]/bg/main.js:265:38
 Could not establish connection. Receiving end does not exist. collectSeen@moz-extension://[NoScript's UUID]/bg/main.js:265:38
  log.js:12:5
	error moz-extension://[NoScript's UUID]/lib/log.js:12:5
	collectSeen moz-extension://[NoScript's UUID]/bg/main.js:274:10
NS_ERROR_NOT_AVAILABLE: Component returned failure code: 0x80040111 (NS_ERROR_NOT_AVAILABLE) [nsIWebNavigation.loadURIWithOptions]  browser-child.js:359
    • Trac
      Trac @tracbot ·

      You can easily observe the increase in extension initialization time with different security slider settings by disabling and enabling extensions in about:addons. During this initialization time the browser UI is unresponsive.

      In normal usage (on my machine anyway) this delay mostly lengthens the time during startup between Tor connecting and the browser window appearing.

      With the security slider set to standard, uBlock Origin with most of the default filters enabled initializes after about 5 seconds of one cpu core pegged at 100%. With the security slider set to safer, it takes 25 seconds.

      I testd with Tor Browser 8.0, Windows 10, a quad-core 3 GHz CPU and plenty of free RAM. Hardware acceleration is disabled and content processes are reduced to 1 although raising them to 2 did not change the browser startup delay or the unresponsive UI in the disable / enable scenario.

      Trac:
      Username: cypherpunks_reply

    • Tom Ritter
      Tom Ritter @tjr ·

      the flag for enabling wasm is on the JSContext

      ​https://searchfox.org/mozilla-central/rev/c0ca77697c6868482f30af873ec8069f2c080a34/js/public/ContextOptions.h#24

      We would need to push it down to CompartmentOptions. And then we can enable it per-Compartment, and enable it for Chrome but not Content.

    • Tom Ritter
      Tom Ritter @tjr ·

      The wasm aspect of this work is happening in https://bugzilla.mozilla.org/show_bug.cgi?id=1576254

    • Tom Ritter
      Tom Ritter @tjr ·

      That bug now has an ESR68 backport of the patch

    • Georg Koppen
      Georg Koppen @gk ·

      Let's try testing that in 9.5a2.

      Trac:
      Keywords: N/A deleted, GeorgKoppen201910, TorBrowserTeam201910 added
      Status: needs_information to new

    • Georg Koppen
      Georg Koppen @gk ·

      Moving my tickets.

      Trac:
      Keywords: GeorgKoppen201910 deleted, GeorgKoppen201911 added

    • Pili Guerra
      Pili Guerra @pili ·

      Moving tickets to November 2019

      Trac:
      Keywords: TorBrowserTeam201910 deleted, TorBrowserTeam201911 added

    • Pili Guerra
      Pili Guerra @pili ·

      Trac:
      Points: N/A to 1

    • A
      Alex Catarineu @acat ·

      Applied tom's backported patches: https://github.com/acatarineu/tor-browser/commits/23719 and did a pass of clang-format that fixed some minor style issues.

      Trac:
      Keywords: TorBrowserTeam201911 deleted, TorBrowserTeam201911R added
      Status: new to needs_review
      Actualpoints: N/A to 0.5

    • Georg Koppen
      Georg Koppen @gk ·

      Do we have a good test making sure the WASM part is actually working as expected?

    • Loading
    • Loading
    • Loading
    • Loading
    • Loading
    • Loading
    • Loading
    • Loading
    • Loading
    • Loading
    Please register or sign in to reply