RE: Windows Vista Power Management & Local Security Policy
As stated in my original e-mail to the list, I definitely don't think that
this is a security vulnerability in a traditional sense. I completely agree
with you. Think about it this way... When you press the power button on the
machine and it performs a graceful shutdown, stuff happens inside of the
operating system. That stuff happens at an elevated privilege level. If
there were some way to hook into the stuff that happens, you (as an
unauthenticated user), could do bad things (besides simply shutting down the
system) using that hook simply by pressing the power button at the logon
screen. For example, if Jim wants to know what Nancy is working on, he could
write a program which e-mails him the contents of her "My Documents" folder
that is triggered by a hook into that process. All Jim needs to do is get
Nancy to run that program on her system (not hard) and walk by her office
when she's not there and hit the power button (also not hard). So what can
_I_ do with this bug? Not much, I'm not that great of a programmer... but I
think someone out there could do some nasty stuff.
--
Abe Getchell
me@xxxxxxxxxxxxxxx
https://abegetchell.com/
> -----Original Message-----
> From: Jim Harrison [mailto:Jim@xxxxxxxxxxxx]
> Sent: Saturday, July 19, 2008 1:36 AM
> To: 'me@xxxxxxxxxxxxxxx'; bugtraq@xxxxxxxxxxxxxxxxx
> Subject: RE: Windows Vista Power Management & Local Security Policy
>
> Abe,
>
> Other than a denial-of-service from the console (is the power switch
> now a security vuln, too?), what can you do with this bug? It's
> absolutely, unquestionably a "bug"; the user should see behavior as
> dictated by logic and described in the documentation, but a "security
> vulnerability"?
>
> I think that's stretching things juuuuuust a bit.
>
> Jim
>
> -----Original Message-----
> From: Abe Getchell [mailto:me@xxxxxxxxxxxxxxx]
> Sent: Thursday, July 17, 2008 7:39 PM
> To: bugtraq@xxxxxxxxxxxxxxxxx
> Subject: Windows Vista Power Management & Local Security Policy
>
> When the security option "Shutdown: Allow system to be shutdown without
> having to log on" (in the local security policy) is set to "Disable",
> and
> the power management setting "When I press the power button" is set to
> "Shut
> Down", it is possible for an unauthenticated user to press the power
> button
> at the Windows logon screen and gracefully shutdown the system. The
> explanation of this security option, taken from the local security
> policy,
> is as follows:
>
> "Shutdown: Allow system to be shut down without having to log on
>
> This security setting determines whether a computer can be shut down
> without
> having to log on to Windows.
>
> When this policy is enabled, the Shut Down command is available on the
> Windows logon screen.
>
> When this policy is disabled, the option to shut down the computer does
> not
> appear on the Windows logon screen. In this case, *users must be able
> to log
> on to the computer successfully and have the Shut down the system user
> right
> before they can perform a system shutdown*.
>
> Default on workstations: Enabled.
> Default on servers: Disabled."
>
> Note the text between the asterisks. While this bug isn't necessarily a
> software flaw allowing for an intrusion into the system in a
> traditional
> sense, it does set a bad precedence in that power management has a free
> pass
> to bypass local security policy and perform actions expressly against
> the
> defined policy. It appears that the only impact the use of this
> security
> option actually has is enabling or disabling the display of the "power
> button" on the Windows logon screen (locally only - this setting has no
> affect on remote desktop connections - the "power button" is not
> displayed
> in either case), not actually preventing anyone from (gracefully)
> shutting
> down the system without logging in.
>
> I reported this to the MSRC on 6/25/2008 and their stance was that this
> wasn't a security vulnerability, but was likely a bug, and was passed
> directly to the product team to investigate through their normal bug
> triage
> process. After some back and forth, there was silence, and I let them
> know I
> was going to release this information to the community.
>
> This was tested on Windows Vista SP1 (32-bit).
>
> --
> Abe Getchell
> me@xxxxxxxxxxxxxxx
> https://abegetchell.com/
>
>