Closed Bug 675804 Opened 9 years ago Closed 9 years ago

e10s HTTP redirects - test_bug397427 fails in Fennec

Categories

(Core :: Networking: HTTP, defect)

x86
Windows 7
defect
Not set

Tracking

()

RESOLVED FIXED
mozilla8

People

(Reporter: martijn.martijn, Assigned: mayhemer)

References

()

Details

(Keywords: testcase)

Attachments

(1 file)

See bug 397427 of why I made this bug security sensitive.
Fennec is failing some of the tests in test_bug397427.html, making it susceptible to those attacks, afaict. (although the description in bug 397427 sounds rather vague  to me, and since Fennec hasn't gained major marketshare, I would think this bug can be unhidden).

See testcase here:
http://www.kantjils.nl/moz/mochitestjs/select.html

It should say (and Firefox desktop does):
style.sheet.cssRules[0].styleSheet.href: http://www.kantjils.nl/moz/mochitestjs/bodyred.css
style.sheet.cssRules[1].styleSheet.href: http://www.kantjils.nl/moz/mochitestjs/redirect.php?bodygreen.css
c.href:c.sheet.href http://www.kantjils.nl/moz/mochitestjs/redirect.php?bodygreen.css:http://www.kantjils.nl/moz/mochitestjs/redirect.php?bodygreen.css

But in Fennec it does say:
style.sheet.cssRules[0].styleSheet.href: http://www.kantjils.nl/moz/mochitestjs/bodyred.css
style.sheet.cssRules[1].styleSheet.href: http://www.kantjils.nl/moz/mochitestjs/bodygreen.css
c.href:c.sheet.href http://www.kantjils.nl/moz/mochitestjs/redirect.php?bodygreen.css:http://www.kantjils.nl/moz/mochitestjs/bodygreen.css

Here is the original mochitest, btw, which partly fail in Fennec:
http://www.kantjils.nl/moz/mochitestjs/test_bug397427.html
We badly set mOriginalURI on redirected child channel on the content process.  I will check why, probably my fault.

When exactly this started to appear?  Wasn't it always failing?
Attached patch v1Splinter Review
Not sure why I have introduced the new member, but it is apparently always left null.
Assignee: nobody → honzab.moz
Status: NEW → ASSIGNED
Attachment #550182 - Flags: review?(jduell.mcbugs)
This is probably one of the many tests we haven't been running. I'm not even certain that it's in the tests we're running right now.
Yes, the plan is to enable the layout/style/test/ mochitests, but we need to figure out which ones are failing, first. And then fix the tests or disable them otherwise.
Attachment #550182 - Flags: review?(jduell.mcbugs) → review+
Keywords: checkin-needed
Would this also fix this case? http://www.kantjils.nl/moz/mochitestjs/body_onload_script_redirect.html
That page never stops loading in Fennec, it works fine in Firefox.

This is probably the cause for failures in:
http://mxr.mozilla.org/mozilla-central/source/layout/style/test/test_visited_image_loading.html?force=1
(In reply to comment #5)
> Would this also fix this case?
> http://www.kantjils.nl/moz/mochitestjs/body_onload_script_redirect.html
> That page never stops loading in Fennec, it works fine in Firefox.
> 

The patch has no affect on this issue.

> This is probably the cause for failures in:
> http://mxr.mozilla.org/mozilla-central/source/layout/style/test/
> test_visited_image_loading.html?force=1

I can see "ASSERTION: Redirecting to a protocol that doesn't support universal protocol redirect"  when running the test.  If that is cause of the test failure then it is quit different issue.  See https://bugzilla.mozilla.org/show_bug.cgi?id=661604#c2 that might be related.

However, we should track all "universal protocol redirect" failing tests and fix the code ; in a different bug.
Yes, bug 661604 sounds exactly what http://www.kantjils.nl/moz/mochitestjs/body_onload_script_redirect.html is suffering from in Fennec.
http://hg.mozilla.org/mozilla-central/rev/e6ea47d9134e
Status: ASSIGNED → RESOLVED
Closed: 9 years ago
Resolution: --- → FIXED
Whiteboard: [inbound]
Target Milestone: --- → mozilla8
Group: core-security
You need to log in before you can comment on or make changes to this bug.