Set referrer policy default to strict-origin-when-cross-origin
Categories
(Core :: DOM: Security, enhancement, P2)
Tracking
()
Tracking | Status | |
---|---|---|
firefox87 | --- | fixed |
People
(Reporter: tnguyen, Assigned: dimi)
References
Details
(Keywords: dev-doc-complete, privacy, Whiteboard: [domsecurity-active])
Attachments
(13 files, 2 obsolete files)
47 bytes,
text/x-phabricator-request
|
Details | Review | |
47 bytes,
text/x-phabricator-request
|
Details | Review | |
47 bytes,
text/x-phabricator-request
|
Details | Review | |
47 bytes,
text/x-phabricator-request
|
Details | Review | |
47 bytes,
text/x-phabricator-request
|
Details | Review | |
47 bytes,
text/x-phabricator-request
|
Details | Review | |
47 bytes,
text/x-phabricator-request
|
Details | Review | |
47 bytes,
text/x-phabricator-request
|
Details | Review | |
47 bytes,
text/x-phabricator-request
|
Details | Review | |
47 bytes,
text/x-phabricator-request
|
Details | Review | |
47 bytes,
text/x-phabricator-request
|
Details | Review | |
47 bytes,
text/x-phabricator-request
|
Details | Review | |
48 bytes,
text/x-phabricator-request
|
Details | Review |
https://github.com/w3c/webappsec-referrer-policy/pull/125
And Google planned to ship that, I think it's worth shipping it to in Firefox. But maybe we can wait until successfully shipping in Chrome
Reporter | ||
Updated•5 years ago
|
Comment 1•5 years ago
|
||
FYI: We already set this strict-origin-when-cross-origin for known trackers in Nightly according to bug 1533763.
Comment 3•5 years ago
|
||
(In reply to Frederik Braun [:freddyb] from comment #1)
FYI: We already set this strict-origin-when-cross-origin for known trackers in Nightly according to bug 1533763.
Yes, but much more broadly, we've been shipping strict-origin-when-cross-origin as the default referrer policy for private browsing mode for several years now.
Reporter | ||
Updated•5 years ago
|
Reporter | ||
Updated•5 years ago
|
Reporter | ||
Comment 5•5 years ago
|
||
Comment 6•5 years ago
|
||
It is probably a good idea to measure the amount of unsafe-url usage using a long-term telemetry probe to make sure websites aren't reverting to using that en masse in response to this change.
Comment 7•5 years ago
|
||
We could try to ship https://github.com/w3c/webappsec-referrer-policy/pull/124 at the same time.
Comment 8•5 years ago
|
||
(In reply to Anne (:annevk) from comment #7)
We could try to ship https://github.com/w3c/webappsec-referrer-policy/pull/124 at the same time.
Interesting idea!
Comment 9•5 years ago
|
||
Anne, I think this one would be good to get out the door, right? Or can it remain in the backlog for now?
Comment 10•5 years ago
|
||
Yeah, it would be good to do this (together with unsafe-url -> no-referrer-when-downgrade).
Comment 11•5 years ago
|
||
Comment 12•5 years ago
|
||
Chrome Intent to Ship: New referrer policy default of strict-origin-when-cross-origin
https://groups.google.com/u/1/a/chromium.org/g/blink-dev/c/lqFuqwZDDR8/m/nKPRTc7DAQAJ
Comment 13•5 years ago
|
||
Bug 970136 seems a duplicate
Assignee | ||
Updated•5 years ago
|
Assignee | ||
Comment 14•4 years ago
|
||
Assignee | ||
Comment 15•4 years ago
|
||
Depends on D88547
Assignee | ||
Comment 16•4 years ago
|
||
Depends on D88548
Assignee | ||
Comment 17•4 years ago
|
||
Depends on D88549
Assignee | ||
Comment 18•4 years ago
|
||
Depends on D88550
Assignee | ||
Comment 19•4 years ago
|
||
Depends on D88551
Assignee | ||
Comment 20•4 years ago
|
||
Depends on D88552
Assignee | ||
Comment 21•4 years ago
|
||
Depends on D88553
Assignee | ||
Comment 22•4 years ago
|
||
Depends on D88554
Assignee | ||
Comment 23•4 years ago
|
||
Depends on D88555
Assignee | ||
Comment 24•4 years ago
|
||
Depends on D88556
Updated•4 years ago
|
Comment 25•4 years ago
|
||
Chrome shipped this so we could follow, I think. WPT has also been updated.
Assignee | ||
Comment 26•4 years ago
|
||
(In reply to Anne (:annevk) from comment #25)
Chrome shipped this so we could follow, I think. WPT has also been updated.
Thank you for the information. I'll update the patch and make sure we pass all those testcases.
Updated•4 years ago
|
Updated•4 years ago
|
Assignee | ||
Comment 27•4 years ago
|
||
We now changes the default referrer policy to strict-origin-when-cross-origin to follow the spec.
This patch reverts the changes made related to referrer policy in Bug 1678042 (bypassing referrer failures after syncing latest wpt tests).
Depends on D88555
Assignee | ||
Comment 28•4 years ago
|
||
Depends on D98435
Assignee | ||
Comment 29•4 years ago
•
|
||
Hi Anne, Christoph and Steven
The current plan is to land this change in 86 cycle, but there are two questions I would like to check with you before landing this:
- Do we want to implement Comment 7 (make'unsafe-url' an alias for 'no-referrer-when-downgrade') along with the default policy change?
- Do we also want to allow websites to override the default with a looser policy, or we prefer still allowing it at this moment and maybe add a telemetry to see how often sites will change it in the future?
Thanks for your help!
Comment 30•4 years ago
|
||
That's great to hear Dimi. To address your questions in order:
- I think we should not touch
unsafe-url
for now as I don't think Chrome has either, but I pinged the PR to see if there are things we might be missing there. (Note that your link has an additional 0 at the end that causes it to 404.) - I think for now we want to allow websites to override this, similar to Chrome, but longer term we should indeed be thinking about clamping that down. Telemetry and analysis of websites would definitely help. The risk with further clamping down is that websites will still leak this state via other means that might be harder to catch so we need to carefully think through that tradeoff as well.
Comment 31•4 years ago
|
||
I suspect you meant to ping me Dimi. Anne's comment makes sense to me.
Assignee | ||
Comment 32•4 years ago
|
||
Hi Anne, Steven,
The original plan was to land this patch in the early 86 cycle, but after discussing with christoph, we both agree that landing the change right before Christmas may not be a good idea. We prefer changing the plan to land it in the early of 87 (25. Jan), so we will have time to react to breakages, if any. But we are also fine to land it when the holiday is over (3.Jan) if 87 is too late. Please let me know if you have any thoughts on this.
Comment 33•4 years ago
|
||
Seems fine, we're not a in rush. (I'll also clear Steven's needinfo since I very much doubt he'd disagree.)
Comment 34•4 years ago
|
||
Comment 35•4 years ago
|
||
Backed out for causing failures on test_DownloadCore.js.
Backout link: https://hg.mozilla.org/integration/autoland/rev/ba10dda56e5e54b6d644e2dac1e41ca926d30e84
Failure log: https://treeherder.mozilla.org/logviewer?job_id=327688109&repo=autoland&lineNumber=7162
Assignee | ||
Comment 36•4 years ago
|
||
Comment 37•4 years ago
|
||
Comment 38•4 years ago
|
||
Comment 39•4 years ago
|
||
Comment 40•4 years ago
|
||
bugherder |
https://hg.mozilla.org/mozilla-central/rev/414252d2e46c
https://hg.mozilla.org/mozilla-central/rev/4d7a026e765e
https://hg.mozilla.org/mozilla-central/rev/695ee14cd948
https://hg.mozilla.org/mozilla-central/rev/9831c19f47b9
https://hg.mozilla.org/mozilla-central/rev/c1b832f3ce06
https://hg.mozilla.org/mozilla-central/rev/dd516f0ffc30
https://hg.mozilla.org/mozilla-central/rev/be5da5f62651
https://hg.mozilla.org/mozilla-central/rev/c0d300b33c6b
https://hg.mozilla.org/mozilla-central/rev/87753b6ce36f
https://hg.mozilla.org/mozilla-central/rev/05c83764972e
https://hg.mozilla.org/mozilla-central/rev/cbe882c7bff5
https://hg.mozilla.org/mozilla-central/rev/0b948812b858
https://hg.mozilla.org/mozilla-central/rev/9a469bfcf4f9
https://hg.mozilla.org/mozilla-central/rev/328001ac7573
https://hg.mozilla.org/mozilla-central/rev/9a5186247b7a
https://hg.mozilla.org/mozilla-central/rev/e1bfeb2c467c
https://hg.mozilla.org/mozilla-central/rev/69dfc7e086ec
https://hg.mozilla.org/mozilla-central/rev/1bdccdb8e90b
https://hg.mozilla.org/mozilla-central/rev/aa07696c3f85
https://hg.mozilla.org/mozilla-central/rev/36db07f697fd
https://hg.mozilla.org/mozilla-central/rev/2e54315e8428
https://hg.mozilla.org/mozilla-central/rev/5555815ae0a6
https://hg.mozilla.org/mozilla-central/rev/2c6b1318917b
Assignee | ||
Updated•4 years ago
|
Comment 41•4 years ago
|
||
FYI the MDN documentation for this in FF87 is (very nearly) complete and can be tracked here. The "live" docs here indicate the new preference. The browser compatibility update with relevant versions should merge in the next couple of days. Please comment in https://github.com/mdn/content/issues/2516 if you think we're missing anything obvious.
Comment 42•4 years ago
|
||
Documentation done, so marking as DDC. Thanks for the great work Hamish!
Comment 43•4 years ago
|
||
when i updated to ff87 my network.http.referer.XOriginTrimmingPolicy (which was on default setting 0) did not change. is this only for new installations and refreshes? and if so is it acceptable to open an issue asking for setting the new default also on updates (perhaps for cases the user did not change the value)?
Comment 44•4 years ago
|
||
(In reply to eyal gruss (eyaler) from comment #43)
when i updated to ff87 my network.http.referer.XOriginTrimmingPolicy (which was on default setting 0) did not change...
The pref changed here was network.http.referer.defaultPolicy
- https://hg.mozilla.org/mozilla-central/rev/4d7a026e765e
Description
•