Although I wasn’t really talking about the technical merit or lack there
of your proposal, with the full details, it’s even worse advice.
First, the checking the registry does not help you. Given that he wants
to write a native application, plug and play could render that value
meaningless, there are well known ways to load a driver that doesn’t
have a registry entry, and it still suffers from the same basic problem
that is unavoidable - it could all be compromised, falsified, et. c.,
and if it’s ahead of you by any of those means, you’re still not going
to be able to boot.
I agree with 100% that there are indeed situations where you have to do
the best you can. I said this a few posts ago, but I also qualified it
as there being nothing ‘generic’ about it, and also that it had to be
profitable to do so. What he would be attempting would be more
‘unknown’ than ‘generic,’ and it damn sure wouldn’t be profitable for
him or anyone else to do so, under his conditions. Either way, I’m not
sure that I would call what you suggested ‘thinking out the box;’ it’s
just not a good idea.
Similarly, the only thing that makes AV’s viable is that they are
profitable, because otherwise, they most certainly do not do good things
to system stability, they don’t exactly feature a high success rate,
which is what it is on a server, but on the desktop, given all that, the
recurring direct cost, and the huge indirect costs from having to
replace otherwise serviceable with new machines just to run an A/V, and
that AV are subject to exactly the same moving target problem, but on
steroids, they make no sense to me.
Incidentally, how this guy’s AV working out for him?
None of these are the major reason why I think your advice is awful.
Let’s assume that there are no other problems with the feasibility of
what you’re proposing. Nevertheless, if I understand your correctly,
you’re suggesting that someone who appears to know close to nothing
about windows drivers, reverse engineer the registry, develop a driver
based on this information, learn how to use windbg, debug it, then
deploy it to 2500 machines around the world, where he can’t do anything
about the problems that it will create, except reinstall the os? You’re
saying that is a more profitable solution that reinstalling the OS?
That’s why I think it’s terrible advice, and that you continue to feed
it to him is really not helping him out, in my opinion.
mm
Deepak Gupta wrote:
On Mon, Jan 19, 2009 at 9:06 PM, Martin O’Brien
> wrote:
>
> This is some of the worst advice I’ve ever seen posted here. As Don
> already mentioned, if the system is configured to ‘require’ the
> file, you’re not going to be able to boot.
>
>
> If you are going to delete a file, then you can also check the registry
> configuration too for checking if it is a boot driver or not. Off course
> you need to reverse engineer the registry which changes from one Windows
> version to another.
> Well I would say this reverse engineering thing only makes Anti Virus
> Products viable in market.
> You can’t fight a modern thief with all old rules, some times you also
> need to break some rules by making sure that it will not break the system.
>
>
>
> I’ll give it one last try - the only thing that you, I or anyone
> else can safely do in a situation like this is reinstall Windows.
> Even if you wrote something that removed the particular file
> safely, once a system has been compromised, there really is no
> reasonable way (if at all, depending on how you analyze it) to
> determine the extent to which it has been compromised, and if you
> include data in this evaluation, there’s no way. In addition, I
> don’t mean this critically, but you clearly are new to all of this,
> and as with everyone who’s new, you don’t know what you’re doing, so
> you really and truly have absolutely no chance of getting this to
> work in a profitable time frame. Whatever time you envision saving
> with this method v. reinstalling the 2500 machines will not
> materialize, and what you will end up with is 2500 machines that you
> have to debug remotely, and I’m guessing you don’t know how to use
> windbg, which in and of itself, is a total deal breaker, and would
> keep you busy for at least the next three months, and in the end,
> you’re still going to have to reinstall windows everywhere.
>
> mm
>
>
> Deepak Gupta wrote:
>
> I don’t know whether this malware is sitting into your File
> System stack or storage stack or not.
>
> But one solution is not delete the file. Just try to corrupt it
> by obfuscating it’s PE header.
> To do this, try doing a raw disk read/write (again it has been
> discussed many times on the list as a dangerous practice)
> Obviously it will be a riskier job to do on end user’s machine
> (So before corrupting it, take a back up of it at your
> quarantine location)
>
> After you have corrupted it, Windows will not load it and on
> next boot you can delete it.
>
> Regards
> Deepak
>
> On Mon, Jan 19, 2009 at 3:31 PM, Pavel A. > mailto:xxxxx mailto:xxxxx> mailto:xxxxx>> wrote:
>
> Maxim S. Shatskih wrote:
>
> The problem is do it in 2500 computers
>
>
> Booting each machine off proper CD is the simplest
> possible thing.
>
> Preparing a proper WinPE CD with a proper script is 10-20
> times
> easier then develop the filter driver, test it and deploy on
> 2500 machines.
>
>
> Can it be done this way:
> the native executable or driver reboots from the
> WinPE/BartPE/Linux CD
> (after checking that it is the correct CD)
> then the cleanup app that runs from the CD will remove the
> former,
> so the next boot from the hard disk will go as usual.
>
> The procedure will be as easy for the users as possible:
>
> 1. Deliver the CD to users, or instruct them to burn the CD.
> 2. Instruct the user to install something from that CD
> ( this will cause reboot from the cd on the next boot)
> 3. Reboot with the CD inserted
>
> This will spare them from discovering in the BIOS how to boot
> from
> CD, and then change it back (it’s a mentality issue… can’t
> be helped)
>
> Regards,
> --PA
>
>
> —
> NTDEV is sponsored by OSR
>
> For our schedule of WDF, WDM, debugging and other seminars visit:
> http://www.osr.com/seminars
>
> To unsubscribe, visit the List Server section of OSR Online at
> http://www.osronline.com/page.cfm?name=ListServer
>
>
>
> —
> NTDEV is sponsored by OSR
>
> For our schedule of WDF, WDM, debugging and other seminars visit:
> http://www.osr.com/seminars
>
> To unsubscribe, visit the List Server section of OSR Online at
> http://www.osronline.com/page.cfm?name=ListServer
>
></mailto:xxxxx></mailto:xxxxx></mailto:xxxxx>