New issue
Advanced search Search tips

Issue 644437 link

Starred by 3 users

Issue metadata

Status: WontFix
Owner: ----
Closed: Sep 2016
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Linux
Pri: 2
Type: Bug-Security



Sign in to add a comment

Omnibox search leaks internal hostnames

Reported by m...@mail.ru, Sep 6 2016

Issue description

UserAgent: Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:48.0) Gecko/20100101 Firefox/48.0

Steps to reproduce the problem:
1. let suppose you have a some web server on internal network aliased as `my-alias.loc' and there is valid secret URI on this server `/my-secret-uri'
2. now type in location bar `my-alias.loc/my-secret-uri', press enter
3. browser sends the data to external search engine (google in my case)
4. you will see a search results and above them valid URI with a comment: "Did you mean to go to `http://my-alias.loc/my-secret-uri'?"

What is the expected behavior?
it is expected, that browser will at least try to resolve a domain name given, before sending the data to external nets

What went wrong?
Currently it is not even trying to perform name resolution, before sending the data piece to external nets

Did this work before? N/A 

Chrome version: 52.0.2743.116 (Developer Build) Built on Ubuntu , running on Ubuntu 16.04 (64-bit)  Channel: n/a
OS Version: Ubuntu 16.04 LTS
Flash Version: Shockwave Flash 11.2 r202
 

Comment 1 by m...@mail.ru, Sep 6 2016

even worser if your alias is for an externally available server, and uri is actually is not to be disclosed
Components: UI>Browser>Omnibox
Summary: Omnibox search leaks internal hostnames (was: Yet another URL leakage)

Comment 3 by m...@mail.ru, Sep 6 2016

well, URI part does matter too. e.g. to avoid backend website software attacks it is better to have backend uri secure. so testing internally some website backend having such secret uri I do transfer it right to untrusted environment. And if server is actually live one and alias is just for a faster typing then we get an actual threat

Comment 4 by wfh@chromium.org, Sep 7 2016

Cc: rsleevi@chromium.org
Components: Internals>Network>DNS
Owner: davidben@chromium.org
Status: Assigned (was: Unconfirmed)
.loc isn't a top level domain specified by IANA - see http://data.iana.org/TLD/tlds-alpha-by-domain.txt so I don't think Chrome should be doing a resolution here.

Passing to the network folks to confirm.

Comment 5 by m...@mail.ru, Sep 7 2016

please note, that firefox (and, surprisingly, even IE) handles this properly, trying to resolve it regardless of is '.loc' TLD or not. Moreover, anything that is not a TLD must be checked twice, not TLD means that is private, so need to be twice careful
Cc: pkasting@chromium.org davidben@chromium.org
Components: -Internals>Network>DNS
Owner: ----
Status: Untriaged (was: Assigned)
Removing assignment, marking this Untriaged.

I'm fairly certain this is WontFix/WAI. ICANN (via SSAC) has certainly highlighted repeatedly the risks of "TLD squatting", given ICANN's desire to make them widely available, and it does not (and cannot) be guaranteed to be kept private, so organizations that rely on this as a privacy/security feature are already in trouble.

I'll let Peter do the WontFix, he's the primary contact for the Omnibox design.
Status: WontFix (was: Untriaged)
We don't try to resolve names before searching because the vast majority of all searches are, potentially, valid intranet names.  So we would have to resolve almost everything, which could have horrific performance implications for omnibox searches, especially in broken and badly-behaving DNS resolution scenarios.  Furthermore, we could no longer tell people ahead of time what their input will do; today, the omnibox makes it clear via icon what will happen when you hit enter, before you hit enter.

Other browsers either don't rely on the address bar as the primary search box and thus don't care about the potential problems here, or have not thought about them.

Chrome mitigates this by remembering any typed navigation, as well as infobar-triggered navigations, and defaulting to navigating in the future.  If you successfully navigate to my-alias.loc once (try just putting "my-alias.loc/" or "http://my-alias.loc" into the address bar, or clicking the link on the "did you mean" infobar you saw), we won't treat this as a search query again.

If you have secret paths that you don't want to leak, you should probably be accessing things over https to begin with.  Otherwise, even if navigation is the default on hitting enter, you could be leaking the path in search suggestion requests while typing.  Whereas if you start typing "https://my-alias.loc/my-secret-uri", we will stop sending suggest requests when you start typing the path, to avoid leaking information.

Comment 8 by m...@mail.ru, Sep 7 2016

IMHO, at least you can add an option to toggle that on/off
No, we will not be adding an option.
Project Member

Comment 10 by sheriffbot@chromium.org, Dec 15 2016

Labels: -Restrict-View-SecurityTeam allpublic
This bug has been closed for more than 14 weeks. Removing security view restrictions.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

Sign in to add a comment