Draft: Bug 31075&41762: FP-Isolate speculative connections
Merge Info
Related Issues
Backporting
Timeline
-
Immediate: patchset needed as soon as possible -
Next Minor Stable Release: patchset that needs to be verified in nightly before backport -
Eventually: patchset that needs to be verified in alpha before backport -
No Backport (preferred): patchset for the next major stable
(Optional) Justification
-
Emergency security update: patchset fixes CVEs, 0-days, etc -
Censorship event: patchset enables censorship circumvention -
Critical bug-fix: patchset fixes a bug in core-functionality -
Consistency: patchset which would make development easier if it were in both the alpha and release branches; developer tools, build system changes, etc -
Sponsor required: patchset required for sponsor -
Other: please explain
Merging
-
Merge to tor-browser
-!fixups
totor-browser
-specific commits, new features, security backports -
Merge to base-browser
-!fixups
tobase-browser
-specific commits, new features to be shared withmullvad-browser
, and security backports-
NOTE: if your changeset includes patches to both
base-browser
andtor-browser
please clearly label in the change description which commits should be cherry-picked tobase-browser
after merging
-
NOTE: if your changeset includes patches to both
Issue Tracking
-
Link resolved issues with appropriate Release Prep issue for changelog generation
Review
Request Reviewer
-
Request review from an applications developer depending on modified system: -
NOTE: if the MR modifies multiple areas, please
/cc
all the relevant reviewers (since gitlab only allows 1 reviewer) - accessibility : henry
- android : clairehurst, dan
- build system : boklm
- extensions : ma1
- firefox internals (XUL/JS/XPCOM) : ma1
- fonts : pierov
- frontend (implementation) : henry
- frontend (review) : donuts, richard
- localization : henry, pierov
- macos : clairehurst, dan
- nightly builds : boklm
- rebases/release-prep : dan, ma1, pierov, richard
- security : ma1
- signing : boklm, richard
- updater : pierov
- misc/other : pierov, richard
-
NOTE: if the MR modifies multiple areas, please
Change Description
We've long had a patch that disabled speculative connections (= connections created by the parent before the content process requests them, according to the comment in the code).
The reason is that we can see that connections are actually created on the catch-all circuit or in the wrong FPD.
However, from what I understood, the speculative connection is not then used for the real requests, at least not if it doesn't match the attributes.
Still, it was right to disable it, especially for the "open a link to a new tab" case, in which it would theoretically create a connection to the "new" domain with the old circuit (even though it was possibly not used).
In practice, this might not happen, or might be disabled on the proxy (bad for MB!). Doing a proper test is difficult, because the connection will be open anyway. I've tried with a Python HTTP server, and I didn't get two requests. So, the connection might really be a TCP connection to find it already open.
Anyway, we know how to do things properly, so let's do it.
If we think this is better than the previous state, we could try uplift it very soon, to get feedback from someone who knows more about it (even though it's quite difficult to find that someone
How Tested
First, I tried to see what we got when removing the old patch with the browser.tordomainisolator.loglevel
set to all
.
It will expose all the requests.
At that point, you'll be able to see we get requests through the --unknown--
username, that is the catch-all circuit.
Another thing you might see is that opening links in the new tab uses a connection that is tied to the opener domain, e.g.,
Requested https://example.net/ via example.org:0:...
Then, apply the new patch. No more catch all circuit, and no more links opened with the previous domain.