Bug 31098 – [XSSAuditor] Allow scripts and plug-ins from the same origin

• Component: WebCore JavaScript
• Status: RESOLVED FIXED

[Adam Barth]

I got a report today of a false positive with the XSSAuditor involving loading a SWF from a relative URL supplied in a request parameter.  We can eliminate this false positive by always allowing same-origin loads of scripts and plug-ins.  That should be pretty safe.

[Daniel Bates]

Adam, did you want to look into this? Otherwise, I can.

[Adam Barth]

If you could look into this, that would be great.  We want to do something similar to what we do for the base tag.

[Daniel Bates]

Created attachment 42721 [details]
Patch with test case

Since XSSAuditor::canLoadExternalScriptFromSrc, XSSAuditor::canLoadObject, and XSSAuditor::canSetBaseElementURL should all allow same-origin loads, I defined a new method XSSAuditor::isSameOriginResource, as opposed to inlining the same-origin check.

[Adam Barth]

Comment on attachment 42721 [details]
Patch with test case

Precisely.

[WebKit Commit Bot]

Comment on attachment 42721 [details]
Patch with test case

Rejecting patch 42721 from commit-queue.

Failed to run "['WebKitTools/Scripts/run-webkit-tests', '--no-launch-safari', '--quiet', '--exit-after-n-failures=1']" exit_code: 1
Running build-dumprendertree
Running tests from /Users/eseidel/Projects/CommitQueue/LayoutTests
Testing 11577 test cases.
http/tests/security/xssAuditor/object-src-inject.html -> failed

Exiting early after 1 failures. 9065 tests run.
257.04s total testing time

9064 test cases (99%) succeeded
1 test case (<1%) had incorrect layout
5 test cases (<1%) had stderr output

[Adam Barth]

Dan, I think you'll have to land this manually because of the executable bit.

[Daniel Bates]

OK. Will do.
(In reply to comment #6)
> Dan, I think you'll have to land this manually because of the executable bit.

[Daniel Bates]

Committed r50631: <http://trac.webkit.org/changeset/50631>

[Eric Seidel (no email)]

svn-apply bug is bug 27204.