While working with configuration profiles is likely the preferred configuration mechanism moving forward (if one hasn't already invested in another mechanism), the larger problem remains that there are multiple mechanisms for configuring settings such as these, with unclear relationships between those mechanisms and unclear behaviors within some mechanisms. If only configuration profiles were used it seems to be a relatively simple problem, but that's not realistic. We're dealing with systems that might have a profile configured this way, might be managed via Profile Manager, might have multiple conflicting .plist files across the multitude of locations in the preferences hierarchy (all the by host, by user, system wide, managed vs. local, etc. variants), and so on. Any configuration method specific assessment mechanism is ultimately of little value as long as multiple methods of configuration can be used. 

I've previously gone down the road of attempting to understand the hierarchy of .plist files and locations, but eventually realized that the effective precedence hierarchy differs from setting to setting (not all settings that may be configured in a plist file in one location are honored if set by that same plist file in another location, and other differences), and may be dependent on the state of the session lifecycle (i.e., one precedence hierarchy is used at login that makes a managed preference override a user level preference, but if the user can change the setting during that session it may override the managed preference for the duration of that session). Dealing with just plist files proved to be an impractical assessment problem due to the inconsistent behaviors and lack of documentation making it a frustrating trial and error effort.

What we really need is a definitive way to determine the effective configuration of a setting regardless of configuration methods used. Sort of an extended MCXCompositor behavior I suppose. A capability that does this would be great for assessment, but to ensure proper remediation we also need documentation of how all these configuration methods interact so we can choose the most appropriate method and/or get the desired result regardless of method chosen.

Shane Shaffer
G2, Inc.
shane.shaffer@g2-inc.com  


On Mon, Jun 3, 2013 at 10:31 AM, Haynes, Dan <dhaynes@mitre.org> wrote:

Reset!  Thanks for the clarification, this definitely helps out and does sound much better than having to track down configuration files.  With that said, I think configuration profiles are available starting in OSX 10.7?  On previously versions of OSX, I think we will still need to deal with tracking down plist files.  By any chance, is there a reference that maps OSX configuration settings to the corresponding plist files and preference keys?

 

Also, we will take a pass at creating a new test around configuration profiles using the techniques you described below.


Thanks,

Danny

 

From: Josh Wisenbaker [mailto:dubs@apple.com]
Sent: Friday, May 31, 2013 6:24 PM
To: Haynes, Dan
Cc: scap-on-apple-dev@lists.macosforge.org; OVAL


Subject: Re: [SCAP-On-Apple-Dev] Configuration Profiles vs Plist file diddling

 

I think that there needs to be a bit of a mental reset as to how to validate settings on the Mac as checking on-disk files doesn't actually verify much and it really hasn't for some time now. For example let's take a look at the loginwindow settings. The current tests would look at a file to see if the settings are there.

 

On my test Mac I have the test profile that I made installed. Because I'm doing things with policy, it doesn't exist in the plists or domains.

 

server:~ localadmin$ defaults -currentHost read com.apple.loginwindow

{

    TALAppsToRelaunchAtLogin =     (

                {

            BackgroundState = 2;

            BundleID = "com.apple.terminal";

            Hide = 0;

            Path = "/Applications/Utilities/Terminal.app";

        },

                {

            BackgroundState = 2;

            BundleID = "com.apple.finder";

            Hide = 0;

            Path = "/System/Library/CoreServices/Finder.app";

        }

    );

}

 

Lets try another!

 

server:~ localadmin$ defaults read  /Library/Preferences/com.apple.loginwindow.plist 

{

    OptimizerLastRunForBuild = 25364928;

    OptimizerLastRunForSystem = 168297216;

    lastUser = loggedIn;

    lastUserName = localadmin;

}

 

Hmm. Not there either. Maybe just the basic domain?

 

server:~ localadmin$ defaults read com.apple.loginwindow

{

    TALLogoutReason = Restart;

}

 

No. Not there either! Well crap. Logging out though shows me that the settings are applied!

 

I can't really check the mobileconfig files either since they get turned into a binary blob upon installation in /var/db/ConfigurationProfiles/Store

 

So what can I do?

 

Well the first thing is that we can ask if there are profiles installed!

 

server:~ localadmin$ profiles -H

profiles are installed on this system

 

Sweet! But now what?

 

Now we actually have several options, the easiest is to grab the profile data right from the profiles command as XML. You can also get this from system_profiler, and if you are using directory services based policy you can get it from dscl or mcxquerey as well.

 

server:~ localadmin$ sudo profiles -Pv -o Output.xml

profiles: verbose mode ON

profiles: outputPath = Output.xml

There are 1 configuration profiles installed

pass

 

server:~ localadmin$ less Output.xml 

 

SNIP…

 

  <dict>

                                        <key>PayloadContent</key>

                                        <dict>

                                                <key>loginWindowIdleTime</key>

                                                <integer>900</integer>

                                                <key>loginWindowModulePath</key>

                                                <string>/System/Library/Screen Savers/Flurry.saver</string>

                                        </dict>

                                        <key>PayloadDisplayName</key>

                                        <string>Login Window:  Screen Saver Preferences</string>

                                        <key>PayloadIdentifier</key>

                                        <string>com.apple.mdm.server.example.private.a55249a0-ab8d-0130-272f-001c42d3e5bf.alacarte.loginwindow.6e51c520-ab8e-0130-2731-001c42d3e5bf.screensaver</string>

                                        <key>PayloadType</key>

                                        <string>com.apple.screensaver</string>

                                        <key>PayloadUUID</key>

                                        <string>42a5182d-2af5-dfa8-16b8-d3a3a88c37fb</string>

                                        <key>PayloadVersion</key>

                                        <integer>1</integer>

                                </dict>

 

…SNIP

 

As we can see, the only way to tell if this policy has been applied with a modern method like a configuration profile is to check using tools that are configuration profile aware.

 

For settings that use profiles this makes everything much simpler. You can simply ask for the output of the profiles command and then parse it with XPath or what not. You get all the settings in one place and you don't need to run around tracking down config files.

 

Thanks,

Josh

 

-- 

Josh Wisenbaker

Consulting Engineer - Apple U.S. Commercial and Governmental Sales

 

On May 31, 2013, at 11:12 AM, "Haynes, Dan" <dhaynes@mitre.org> wrote:



Thanks for getting this discussion started Josh and Prabhu!

 

As Prahbu mentioned, we have the existing macos-def:plist510_test and the ind-def:xmlfilecontent_test that can support the assessment of plist files (including configuration profiles).  There is also an experimental plist511_test in the OVAL Language Sandbox (https://github.com/OVALProject/Sandbox/blob/master/x-macos-plist-xpath.xsd) which utilizes xpath to reference the preference key of interest. 

 

To help show how these three tests can be used, I have attached sample definitions that check for CCE-28300-2 idle_time_for_screen_saver.

 

Hope this helps!


Thanks,

Danny

 

From: scap-on-apple-dev-bounces@lists.macosforge.org [mailto:scap-on-apple-dev-bounces@lists.macosforge.org] On Behalf OfPrabhu S Angadi
Sent: Friday, May 31, 2013 3:38 AM
To: scap-on-apple-dev@lists.macosforge.org
Subject: Re: [SCAP-On-Apple-Dev] Configuration Profiles vs Plist file diddling

 

Hi All,

Yes! I completely agree with Josh, on the usage configuration profiles.

Being the XML formatted content of these files can be easily parsed to fetch the composed policies values, to develop the 
SCAP OVALl definitions, using available '< xmlfilecontent_test >' or '< plist510_test >' probes for better assessment.

And also, as these files can be easily deployed with customized values as per user's choice. Either by 

*    By physically connecting the device
*    In an email message
*    On a webpage
*    Using over-the air configuration as described in this document

so I think it will be of great use in remediation part as well.

_______________________________________________________________________________________

In supportive to Josh, I have attached few Profile files, that were developed to address the Apple iOS Hardening Checklists 
by The University Of Texas at Austin.

FMI : 

https://wikis.utexas.edu/display/ISO/Apple+iOS+Hardening+Checklist

https://wikis.utexas.edu/display/ISO/iOS+Configuration+Profiles


-- 
Thanks !! 
Prabhu S A 

http://www.scaprepo.com


On 05/31/2013 02:50 AM, Josh Wisenbaker wrote:

Hi all,

 

I think that from an audit and remediation standpoint things can be greatly simplified by using Configuration Profiles.

 

You can easily get a XML formatted list of the composited policies that are on the Mac and you can easily apply settings by installing a profile. Using the policy mechanisms in OS X is highly recommended over messing with files.

 

As an example here is a profile I made that implements all of the settings for the initial loginwindow tickets that are in the tracker.

 




 

This profile allows for removal without authentication so it's easy to test with.

 

Thoughts?

Josh

 

-- 

Josh Wisenbaker

Consulting Engineer - Apple U.S. Commercial and Governmental Sales

 





_______________________________________________
SCAP-On-Apple-Dev mailing list
SCAP-On-Apple-Dev@lists.macosforge.org
https://lists.macosforge.org/mailman/listinfo/scap-on-apple-dev

<plist510_test_example.xml><plist511_test_example.xml><xmlfilecontent_test_example.xml>_______________________________________________
SCAP-On-Apple-Dev mailing list
SCAP-On-Apple-Dev@lists.macosforge.org
https://lists.macosforge.org/mailman/listinfo/scap-on-apple-dev

 


_______________________________________________
SCAP-On-Apple-Dev mailing list
SCAP-On-Apple-Dev@lists.macosforge.org
https://lists.macosforge.org/mailman/listinfo/scap-on-apple-dev