834836 - Turn on pref to block mixed active content

Status: VERIFIED FIXED

(Firefox :: Security, defect)

Description

[Tanvi Vyas[:tanvi]]

Attachment: Set "security.mixed_content.block_active_content" to true

Set "security.mixed_content.block_active_content" to true by default (instead of false). 

Trying to do this on try shows lots of oranges because of tests that expect the mixed content to load (mixed content tests, onsecurity change tests, csp test times out, local storage test, web console mixed content entry test, etc).  This bug is to update these tests and turn the pref on.

https://tbpl.mozilla.org/?tree=Try&rev=7ff3f12543ac

Comment 1

[Daniel Veditz [:dveditz]]

Comment on attachment 706531 [details] [diff] [review]
Set "security.mixed_content.block_active_content" to true

Review of attachment 706531 [details] [diff] [review]:
-----------------------------------------------------------------

r=dveditz on this patch when you're ready to land (i.e. regressions/broken tests are fixed)

Comment 2

[Tanvi Vyas[:tanvi]]

Attachment: Updated /dom tests so that they don't break when we block mixed active content v1

Comment 3

[Tanvi Vyas[:tanvi]]

Attachment: Updated CSP test so that it doesn't break when we block mixed active content v1

Comment 4

[Tanvi Vyas[:tanvi]]

Attachment: Updated browser_discovery test since it does it's own mixed content handling v1

Comment 5

[Tanvi Vyas[:tanvi]]

Next I have to update all the actual mixed content tests.

Comment 6

[Tanvi Vyas[:tanvi]]

Comment on attachment 712016 [details]
Updated browser_discovery test since it does it's own mixed content handling v1

Spoke to Dave on IRC and decided that the 2 insecure content tests in browser_discovery should be run with the flag set to false and the flag set to true.  Will update the patch.

Comment 7

[Sid Stamm [:geekboy or :sstamm]]

Comment on attachment 712013 [details] [diff] [review]
Updated CSP test so that it doesn't break when we block mixed active content v1

Review of attachment 712013 [details] [diff] [review]:
-----------------------------------------------------------------

Looks good.  r=me on this test tweak.

Comment 8

[Tanvi Vyas[:tanvi]]

Comment on attachment 712011 [details] [diff] [review]
Updated /dom tests so that they don't break when we block mixed active content v1

This patch actually also includes a change to a /browser/devtools test.  So asking additional review from Mihai for just the browser/devtools/webconsole/test/browser_webconsole_bug_737873_mixedcontent.js test.  

Sorry for missing this and not splitting this into separate patches earlier!

Comment 9

[Tanvi Vyas[:tanvi]]

Attachment: Update mixed content tests so that they don't break when we block mixed active content v1

Work in progress patch for the mixed content tests.  I can't seem to run these locally (not sure why??), but have pushed to try - https://tbpl.mozilla.org/?tree=Try&rev=b24426a48ce6

There is one tests that causes some problems when I try to set and then unset the block active and block display tests:
https://mxr.mozilla.org/mozilla-central/source/security/manager/ssl/tests/mochitest/mixedcontent/

This test navigates from one page to another, and hence finish() isn't called on the first page, and the original blockDisplay and blockActive prefs aren't available on the last page.  The value of the prefs actually doesn't matter for that test and hence we can update mixedContentTest.js to ignore it.  Right now, I use the fact that it is the only test that sets bypassNavigationTests to true to determine this.  I can instead use a new flag that is specific to this test if that is preferable.

Feedback? to bsmith and dkeeler.

Comment 10

[Dana Keeler (she/her) (use needinfo) [:keeler] (on leave)]

Comment on attachment 712642 [details] [diff] [review]
Update mixed content tests so that they don't break when we block mixed active content v1

Review of attachment 712642 [details] [diff] [review]:
-----------------------------------------------------------------

Looks like this just caches the values of the mixed content prefs, so if that's all that's needed, then I think this is fine.

::: security/manager/ssl/tests/mochitest/mixedcontent/mixedContentTest.js
@@ +27,5 @@
>  
>  // Internal variables
>  var _windowCount = 0;
> +var blockDisplay;
> +var blockActive;

Maybe call these _savedBlockDisplayPref, _savedBlockActivePref? (underscore for consistency, "saved...Pref" to be a bit more clear about what they're used for)
Also, I would initialize them to null since they get specifically compared with null later.

Comment 11

[Brian Smith (:briansmith, :bsmith, use NEEDINFO?)]

Comment on attachment 706531 [details] [diff] [review]
Set "security.mixed_content.block_active_content" to true

Review of attachment 706531 [details] [diff] [review]:
-----------------------------------------------------------------

::: modules/libpref/src/init/all.js
@@ +1384,5 @@
>  pref("security.csp.enable", true);
>  pref("security.csp.debug", false);
>  
>  // Mixed content blocking
> +pref("security.mixed_content.block_active_content", true);

This will cause mixed active content to be blocked in B2G, Fennec, and Thunderbird too, where there is no UI to unblock it. Is that the desired intent?

Comment 12

[Tanvi Vyas[:tanvi]]

Comment on attachment 706531 [details] [diff] [review]
Set "security.mixed_content.block_active_content" to true

(In reply to Brian Smith (:bsmith) from comment #11)
> Comment on attachment 706531 [details] [diff] [review]
> Set "security.mixed_content.block_active_content" to true
> 
> Review of attachment 706531 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> ::: modules/libpref/src/init/all.js
> @@ +1384,5 @@
> >  pref("security.csp.enable", true);
> >  pref("security.csp.debug", false);
> >  
> >  // Mixed content blocking
> > +pref("security.mixed_content.block_active_content", true);
> 
> This will cause mixed active content to be blocked in B2G, Fennec, and
> Thunderbird too, where there is no UI to unblock it. Is that the desired
> intent?

No.  Thanks for catching this.  I need to move this to firefox.js

Comment 13

[Tanvi Vyas[:tanvi]]

Comment on attachment 712016 [details]
Updated browser_discovery test since it does it's own mixed content handling v1

We need to update the extensions code to detect when Mixed Content is blocked.  This is now being handled in bug 840388.  Obsoleting this patch; the test shouldn't require any changes after all.

Comment 14

[Mihai Sucan [:msucan]]

Comment on attachment 712011 [details] [diff] [review]
Updated /dom tests so that they don't break when we block mixed active content v1

Patch looks good. At the end you can use clearUserPref() to reset back to the default value, instead of setBoolPref(). Also, instead of having a new endTest() function you could registerCleanupFunction(function() { ... clearUserPref() ... }). This is just for future reference - you can land this patch as-is.

Thanks for keeping me in the loop withe web console-related changes.

Comment 15

[Tanvi Vyas[:tanvi]]

(In reply to David Keeler (:keeler) from comment #10)
> Comment on attachment 712642 [details] [diff] [review]
> Update mixed content tests so that they don't break when we block mixed
> active content v1
> 
> Review of attachment 712642 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> Looks like this just caches the values of the mixed content prefs, so if
> that's all that's needed, then I think this is fine.
> 
> ::: security/manager/ssl/tests/mochitest/mixedcontent/mixedContentTest.js
> @@ +27,5 @@
> >  
> >  // Internal variables
> >  var _windowCount = 0;
> > +var blockDisplay;
> > +var blockActive;
> 
> Maybe call these _savedBlockDisplayPref, _savedBlockActivePref? (underscore
> for consistency, "saved...Pref" to be a bit more clear about what they're
> used for)
> Also, I would initialize them to null since they get specifically compared
> with null later.

Looks like I use origBlockActive in other tests, so I will switch the variable name to origBlockActive in all of the tests in this bug too.

Comment 16

[Tanvi Vyas[:tanvi]]

Attachment: Updated /dom and webconsole tests so that they don't break when we block mixed active content v2

Carrying over r+ from Mihai and Olli.

Changed variable names from blockActive and blockDisplay to origBlockActive and origBlockDisplay.

Comment 17

[Tanvi Vyas[:tanvi]]

Attachment: Update mixed content tests so that they don't break when we block mixed active content v2

Update 9 /mixedcontent tests that depend on mixed active content to load in order to test the security state of a page.

security/manager/ssl/tests/mochitest/mixedcontent/test_bug329869.html
security/manager/ssl/tests/mochitest/mixedcontent/test_documentWrite2.html
security/manager/ssl/tests/mochitest/mixedcontent/test_dynDelayedUnsecureXHR.html
security/manager/ssl/tests/mochitest/mixedcontent/test_dynUnsecureIframeRedirect.html
security/manager/ssl/tests/mochitest/mixedcontent/test_unsecureCSS.html
security/manager/ssl/tests/mochitest/mixedcontent/test_unsecureIframe.html
security/manager/ssl/tests/mochitest/mixedcontent/test_unsecureIframe2.html
security/manager/ssl/tests/mochitest/mixedcontent/test_unsecureIframeRedirect.html
security/manager/ssl/tests/mochitest/mixedcontent/test_unsecurePictureInIframe.html
security/manager/ssl/tests/mochitest/mixedcontent/test_unsecureRedirect.html

Comment 18

[Tanvi Vyas[:tanvi]]

Attachment: Set "security.mixed_content.block_active_content" to true for desktop Firefox

Comment 19

[Tanvi Vyas[:tanvi]]

Pushed all of these to try, along with the patch to bug 840388 that handles mixed content on the about:addons page.

https://tbpl.mozilla.org/?tree=Try&rev=ac63fceb7e30

Comment 20

[Brian Smith (:briansmith, :bsmith, use NEEDINFO?)]

Comment on attachment 713063 [details] [diff] [review]
Update mixed content tests so that they don't break when we block mixed active content v2

Tanvi, I think this is the wrong patch. All this stuff is in dom/ and devtools.

Comment 21

[Tanvi Vyas[:tanvi]]

Attachment: Update mixed content tests so that they don't break when we block mixed active content v3

(In reply to Brian Smith (:bsmith) from comment #20)
> Comment on attachment 713063 [details] [diff] [review]
> Update mixed content tests so that they don't break when we block mixed
> active content v2
> 
> Tanvi, I think this is the wrong patch. All this stuff is in dom/ and
> devtools.

Sorry, added the wrong patch.  Updated with the correct patch.

Comment 22

[Brian Smith (:briansmith, :bsmith, use NEEDINFO?)]

Comment on attachment 713132 [details] [diff] [review]
Update mixed content tests so that they don't break when we block mixed active content v3

Review of attachment 713132 [details] [diff] [review]:
-----------------------------------------------------------------

::: security/manager/ssl/tests/mochitest/mixedcontent/mixedContentTest.js
@@ +69,5 @@
>  
> +    if (hasMixedActiveContent)
> +    {
> +      origBlockActive = SpecialPowers.getBoolPref("security.mixed_content.block_active_content")
> +      SpecialPowers.setBoolPref("security.mixed_content.block_active_content", false);

SpecialPowers.pushPrefEnv?

@@ +96,5 @@
>        {
>          if (testCleanUp)
>            testCleanUp();
> +        if (hasMixedActiveContent)
> +          SpecialPowers.setBoolPref("security.mixed_content.block_active_content", origBlockActive);

SpecialPowers.popPrefEnv?

Comment 23

[Tanvi Vyas[:tanvi]]

Attachment: Update mixed content tests so that they don't break when we block mixed active content v4

Updated with bsmith's review comments.  Carrying over r+ and did another push to try, but limited and only re-testing mochitest-5 where the mixed content tests exist:

https://tbpl.mozilla.org/?tree=Try&rev=2cd8832b34c4

Comment 24

[Brian Smith (:briansmith, :bsmith, use NEEDINFO?)]

Comment on attachment 713069 [details] [diff] [review]
Set "security.mixed_content.block_active_content" to true for desktop Firefox

Stealing review from Dan, as Dan already r+d enabling this before.

Comment 25

[Tanvi Vyas[:tanvi]]

Needed a new push to try anyway; cancelling the previous ones: https://tbpl.mozilla.org/?tree=Try&rev=f0a382ea2080

Comment 26

[Justin Dolske [:Dolske]]

Is the intent to land this before the uplift, or at the beginning of the Firefox 22 cycle?

There has been a lot of churn in this feature to get it working, and things like bug 840641 make me think it's not had any significant usage on Nightly (by users manually turning it on). Ergo we don't know if this is likely to cause breakage and need more fixes (which would be undesirable to be doing directly in Aurora).

I don't think this should squeak in at the end of a cycle, it should get as much of a full Nightly cycle as possible.

Comment 27

[Tanvi Vyas[:tanvi]]

Moving target milestone to FF22 because there are still some dependency bugs that need to be fixed.

Comment 28

[Tanvi Vyas[:tanvi]]

Pushed to try, along with the patches in Bug 840388 and Bug 836951:
https://tbpl.mozilla.org/?tree=Try&rev=3b9e02950972

Comment 29

[Tanvi Vyas[:tanvi]]

I took the patches from this bug, bug 836951, bug 840388, and bug 841850 except for
the patch that actually switches the security.mixed_content.block_active_content to true and I pushed them to try.  This try looks good: https://tbpl.mozilla.org/?tree=Try&rev=27e901d5fbe9


I also pushed to try with all of these patches + the pref that switches block_active_content to true.  In this result, there are password manager tests that are failing - https://tbpl.mozilla.org/?tree=Try&rev=d078bae998d3.
/tests/toolkit/components/passwordmgr/test/test_prompt.html 
/tests/toolkit/components/passwordmgr/test/test_notifications_popup.html 

Turning the pref on didn't make these fail before.  But both tests have been recently changed.  I need to dig into what is causing the failures.  Locally,

Comment 30

[Tanvi Vyas[:tanvi]]

(In reply to Tanvi Vyas [:tanvi] from comment #29)
> Turning the pref on didn't make these fail before.  But both tests have been
> recently changed.  I need to dig into what is causing the failures.  Locally,

...

Locally, test_notifications_popup.html completes without a timeout.  But for test_prompt.html, I do get the same error that try is observing:
TypeError: notification is undefined at http://mochi.test:8888/tests/toolkit/components/passwordmgr/test/notification_common.js:43

Comment 31

[Tanvi Vyas[:tanvi]]

The reason that a failure in test http://mxr.mozilla.org/mozilla-central/source/toolkit/components/passwordmgr/test/test_prompt.html?force=1 did not show up before was because the test was disabled and then re-enabled as part of bug 806737.

The failure is caused by a situation where you have a form with a passwords input field and mixed content.  For example, try entering a bogus username and password here: 
https://people.mozilla.com/~tvyas/javascript_post_https.html
When you enter them, you get a doorhanger popup asking if you want Firefox to remember the password.

Now try going to https://people.mozilla.com/~tvyas/javascript_post_https_mixed.html and entering a bogus username and password.  After you hit submit, you will get two icons next to the location bar.  One from the password manager to save the password, and another from the mixed content blocker.  The password manager notification doorhanger is displayed for a split second and then disappears.  I'm not sure why yet.

Justin, Jared - do you have any ideas what is going on here and why the passwords manager doorhanger disappears? The code to show the doorhanger is here http://mxr.mozilla.org/mozilla-central/source/toolkit/components/passwordmgr/nsLoginManagerPrompter.js#828 and shows that the notification should persist for 10 seconds.

I had tested stacked notifications with the click to play notification and the mixed content doorhanger (https://people.mozilla.com/~tvyas/mixedctp.html).  But both of these are dismissed by default.  I hadn't tested it with a page where one doorhanger wants to show up automatically and another doesn't.  Or in the case where both doorhangers want to show up automatically.  What is the correct behavior in these cases?

Thanks!

Comment 32

[Tanvi Vyas[:tanvi]]

This may be an issue with the notification code in general.  The same issue that is occurring with the mixed content blocker notification occurs with the click to play notification.  Go to https://people.mozilla.com/~tvyas/passwordctp.html and enter a bogus username and password.  You will see the remember password doorhanger appear for a split second and then disappear.

Comment 33

[Justin Dolske [:Dolske]]

Yes, you'll either need to fix the tests so they don't trigger the mixed-content notification (and therefore pass), or we'll need to get PopupNotifications.jsm to be smarter about how multiple notifications are handled.

Comment 34

[Tanvi Vyas[:tanvi]]

(In reply to Justin Dolske [:Dolske] from comment #33)
> Yes, you'll either need to fix the tests so they don't trigger the
> mixed-content notification (and therefore pass), or we'll need to get
> PopupNotifications.jsm to be smarter about how multiple notifications are
> handled.

We probably need to do both.  I will file a bug to make PopupNotifications.jsm smarter.  I think the bug is somewhere in the code here: http://mxr.mozilla.org/mozilla-central/source/toolkit/content/PopupNotifications.jsm#554.

Comment 35

[Tanvi Vyas[:tanvi]]

Filed bug 854740 to make PopupNotifications.jsm smarter.  Now I will work on updating the password manager tests so that they don't get multiple notifications.

Comment 36

[Tanvi Vyas[:tanvi]]

The try logs show:
1:20     INFO -  166403 INFO TEST-PASS | /tests/toolkit/components/passwordmgr/test/test_prompt.html | handleLoad running for test 1003
16:51:20     INFO -  166404 INFO TEST-PASS | /tests/toolkit/components/passwordmgr/test/test_prompt.html | Time is Sat, 23 Mar 2013 23:51:20 GMT
16:51:20     INFO -  166405 INFO TEST-PASS | /tests/toolkit/components/passwordmgr/test/test_prompt.html | is popup panel open? false
16:51:20     INFO -  166406 INFO TEST-PASS | /tests/toolkit/components/passwordmgr/test/test_prompt.html | Found 2 popup notifications.
16:51:20     INFO -  166407 INFO TEST-PASS | /tests/toolkit/components/passwordmgr/test/test_prompt.html | #0: mixed-content-blocked
16:51:20     INFO -  166408 INFO TEST-PASS | /tests/toolkit/components/passwordmgr/test/test_prompt.html | #1: password-change
16:51:20     INFO -  166409 INFO TEST-PASS | /tests/toolkit/components/passwordmgr/test/test_prompt.html | Found 0 notification bars.
16:51:20     INFO -  166410 INFO TEST-PASS | /tests/toolkit/components/passwordmgr/test/test_prompt.html | handleDialog was invoked
16:51:20     INFO -  166411 INFO TEST-PASS | /tests/toolkit/components/passwordmgr/test/test_prompt.html | Checking for successful authentication
16:51:20     INFO -  166412 INFO TEST-PASS | /tests/toolkit/components/passwordmgr/test/test_prompt.html | Checking for echoed username
16:51:20     INFO -  166413 INFO TEST-PASS | /tests/toolkit/components/passwordmgr/test/test_prompt.html | Checking for echoed password
16:51:20     INFO -  166414 INFO TEST-PASS | /tests/toolkit/components/passwordmgr/test/test_prompt.html | Looking for password-change popup notification
16:51:20     INFO -  166415 INFO TEST-PASS | /tests/toolkit/components/passwordmgr/test/test_prompt.html | got popup notification
16:51:20     INFO -  166416 INFO TEST-PASS | /tests/toolkit/components/passwordmgr/test/test_prompt.html | Looking for action at index 0
16:51:20     INFO -  166417 ERROR TEST-UNEXPECTED-FAIL | /tests/toolkit/components/passwordmgr/test/test_prompt.html | at least one notification displayed


https://tbpl.mozilla.org/php/getParsedLog.php?id=21023785&tree=Try

Note that is says Found 2 popup notifications.
 #0: mixed-content-blocked
 #1: password-change

This test is an http test and all its subframes are also http, so there is no reason for the mixed-content-blocked popupnotification tos how up.  Running the test locally passes and I don't see any mixed content notifications or doorhangers.

Comment 37

[Tanvi Vyas[:tanvi]]

Pushed to try again: https://tbpl.mozilla.org/?tree=Try&rev=bff4225d1454

Comment 38

[Tanvi Vyas[:tanvi]]

http://hg.mozilla.org/integration/mozilla-inbound/rev/9b46d7f33f26
http://hg.mozilla.org/integration/mozilla-inbound/rev/d2eefc0da377
http://hg.mozilla.org/integration/mozilla-inbound/rev/ae4bb9b986cf

Pushed the mochitest updates to inbound.  Have NOT pushed the patch that enables mixed active content blocking.  We can't do that yet, because it causes the following mochitests to fail:
/tests/toolkit/components/passwordmgr/test/test_prompt.html 
/tests/toolkit/components/passwordmgr/test/test_notifications_popup.html 

Updating the dependency list.  The only blocker remaining to enable the pref is the above password manager tests OR Bug 854740 (not sure yet which of the two are the root of the problem).

Comment 39

[Kohei Yoshino [:kohei]]

I've added this bug to the compatibility doc. Please correct the info if wrong.
https://developer.mozilla.org/en-US/docs/Site_Compatibility_for_Firefox_22

Also, please update relevant docs if any.

Comment 40

[Tanvi Vyas[:tanvi]]

(In reply to Kohei Yoshino from comment #39)
> I've added this bug to the compatibility doc. Please correct the info if
> wrong.
> https://developer.mozilla.org/en-US/docs/Site_Compatibility_for_Firefox_22
> 
> Also, please update relevant docs if any.

We don't know yet if the preference to block mixed active content will be on by default in Firefox 22.  That part of this bug has not landed because of 2 password manager mochitests that are failing.

Comment 41

[Justin Dolske [:Dolske]]

We're in the last few days of the 22 cycle -- I'd have the same concerns about landing at the end of the cycle as I did 6 weeks ago in comment 26.

Comment 42

[Kohei Yoshino [:kohei]]

Commented out this change from the compat doc.

Comment 43

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/9b46d7f33f26
https://hg.mozilla.org/mozilla-central/rev/d2eefc0da377
https://hg.mozilla.org/mozilla-central/rev/ae4bb9b986cf

Comment 44

[Tanvi Vyas[:tanvi]]

In bug 855275, Gavin is working on the PopupNotifications bugs that cause the two password manager tests to fail.  Once that bug is complete, we should be able to turn the pref on.

Unless of course more mochitests are created/enabled in the meantime and they happen to fail with mixed content blocking turned on.  Let's hope we can finish this before that happens again ;)

Comment 45

[Tanvi Vyas[:tanvi]]

Found another test that breaks when mixed content blocking is turned on: 
https://mxr.mozilla.org/mozilla-central/source/dom/browser-element/mochitest/test_browserElement_inproc_SecurityChange.html?force=1

It's looking for a broken security state but blocking mixed content prevents that.  Need to update the test.

Comment 46

[Tanvi Vyas[:tanvi]]

Attachment: Update BrowserElement tests so that they don't break when we block mixed active content

(In reply to Tanvi Vyas [:tanvi] from comment #45)
> Found another test that breaks when mixed content blocking is turned on: 
> https://mxr.mozilla.org/mozilla-central/source/dom/browser-element/mochitest/
> test_browserElement_inproc_SecurityChange.html?force=1
> 
> It's looking for a broken security state but blocking mixed content prevents
> that.  Need to update the test.

Test updated.  r? to smaug (who reviewed the original mochitests in bug 763694.

Comment 47

[Tanvi Vyas[:tanvi]]

Pushed to inbound!!!
https://hg.mozilla.org/integration/mozilla-inbound/rev/4f1779767e3d
https://hg.mozilla.org/integration/mozilla-inbound/rev/60b08f643863

With this push, Mixed Active Content will be blocked by default, with an option to un-block through a doorhanger.

Comment 48

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/4f1779767e3d
https://hg.mozilla.org/mozilla-central/rev/60b08f643863

Comment 49

[Tanvi Vyas[:tanvi]]

Landed in FF 23.

Comment 50

[Kohei Yoshino [:kohei]]

I've added this bug to the site compat doc. Please correct the info if wrong.
https://developer.mozilla.org/en-US/docs/Site_Compatibility_for_Firefox_23

Also, please update relevant docs if any.

Comment 51

[Tanvi Vyas[:tanvi]]

(In reply to Kohei Yoshino from comment #50)
> I've added this bug to the site compat doc. Please correct the info if wrong.
> https://developer.mozilla.org/en-US/docs/Site_Compatibility_for_Firefox_23
> 
> Also, please update relevant docs if any.

There are a couple of corrections, but I don't seem to be able to correct the page.  I keep getting permission denied after logging in, editing, and trying to save.

The title should be:
Non-SSL active content on SSL pages is blocked by default

And the first sentence should read:
Firefox 18 introduced preferences to block loading content from non-SSL (http) sites on SSL (https) pages.

Comment 52

[Jesse Ruderman]

I made the changes you suggested in comment 51.

Can you file a bug about the devmo login/permission problem you're experiencing?

Comment 53

[Kohei Yoshino [:kohei]]

Thanks for your correcting. Tanvi: you have to enable HTTP referer on MDN to save your changes. I have an add-on installed to send referer only when I save. MDN issues a Cookie called "csrftoken" but it also refers HTTP referer to avoid CSRF attacks. Very annoying security practice, IMO.

Comment 54

[Brian Smith (:briansmith, :bsmith, use NEEDINFO?)]

(In reply to Kohei Yoshino from comment #53)
> Thanks for your correcting. Tanvi: you have to enable HTTP referer on MDN to
> save your changes. I have an add-on installed to send referer only when I
> save. MDN issues a Cookie called "csrftoken" but it also refers HTTP referer
> to avoid CSRF attacks. Very annoying security practice, IMO.

Thanks for the info. I filed bug 861237.

Comment 55

[Mihai Morar, (:MihaiMorar)]

I confirm the fix is verified on Windows 7 x64 and Mac OS 10.7.5