1119005 - No need to branch on NewChannel callsites anymore

Status: RESOLVED FIXED

(Core :: DOM: Security, defect)

Description

[Christoph Kerschbaumer [:ckerschb]]

As Jonas pointed out:
https://bugzilla.mozilla.org/show_bug.cgi?id=1110469#c3
we should investigate.

Comment 1

[Christoph Kerschbaumer [:ckerschb]]

Attachment: jarchannel_should_not_pass_system.patch

Jonas, I think we should rather use the nullPrincipal instead of just calling NS_NewChannelInternal(loadInfo) in case there is no loadInfo. We might end up creating a channel with no loadInfo within Gecko. I think we should try to avoid that, no?



In comparison to other cases, which I am going to update soon in a different bug:
http://mxr.mozilla.org/mozilla-central/source/browser/components/about/AboutRedirector.cpp#166

Basically, wherever we left that comment:
> // Bug 1087720 (and Bug 1099296):
> // Once all callsites have been updated to call NewChannel2()
> // instead of NewChannel() we should have a non-null loadInfo
> // consistently. Until then we have to branch on the loadInfo.
we can then just call NS_NewChannelInternal instead of branching on the loadInfo.

Comment 2

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Comment on attachment 8566310 [details] [diff] [review]
jarchannel_should_not_pass_system.patch

Review of attachment 8566310 [details] [diff] [review]:
-----------------------------------------------------------------

If there's no mLoadInfo then the "inner" channel also shouldn't have a loadinfo. So just remove the 'else'.

Comment 3

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

If we don't have a loadinfo then it's no more "correct" to just make up information than to clearly indicate that there's no loadinfo.

Comment 4

[Christoph Kerschbaumer [:ckerschb]]

Attachment: bug_1119005_netutil_changes.patch

Since we allow channel creation without a LoadInfo, and now also allow channel redirection without a loadInfo (bug 1134196), we should finally also allow to create a new InputStreamChannel without a loadInfo.

Comment 5

[Christoph Kerschbaumer [:ckerschb]]

Attachment: bug_1119005_do_not_branch_on_newchannelinternal_callsites.patch

Removing the if condition before calling NewChannelInternal and NewInputStreamChannelInternal since both functions can deal with null loadInfos. Should be a rare case anyway to pass a null loadInfo, since we have update all of the callsites of newChannel within Gecko to provide the necessary arguments.

Comment 6

[Steve Workman [:sworkman] (INACTIVE)]

Comment on attachment 8566685 [details] [diff] [review]
bug_1119005_netutil_changes.patch

Review of attachment 8566685 [details] [diff] [review]:
-----------------------------------------------------------------

Taking Jason's review and marking r+. Looks sane enough to me.

Comment 7

[Christoph Kerschbaumer [:ckerschb]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/9d623faf8c8b
https://hg.mozilla.org/integration/mozilla-inbound/rev/2144c24dfa01

Comment 8

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/9d623faf8c8b
https://hg.mozilla.org/mozilla-central/rev/2144c24dfa01