Archived from groups: alt.comp.hardware.overclocking (
More info?)
David Maynard <dNOTmayn@ev1.net> wrote:
> Mitch Crane wrote:
>> And they aren't always incompatible either. For example, Prime95 may
>> use some programming methodolgy which executes what Windows thinks is
>> data in a perfectly safe and known way--it knows what it's doing even
>> if Windows doesn't--,
>
> You miss the point. Prime may know it's 'safe' but if DEP is forced to
> allow Prime to do peculiar things then it must allow all programs to
> do those peculiar things and the next program may NOT be 'safe'.
Actually, I covered that point later in my post.
> Unless you have the unreasonable expectation that Microsoft will write
> a specific exclusion for the one program called 'Prime' (did anyone
> ASK then to? How would they even know?) and deny the others.
Firewalls have exclusions, for example.
>> because speed is of prime (no pun intended)
>> importance. Your firewall may block certain network operations,
>> because they can be dangerous, but it doesn't block them all. The
>> fact that an app access the net doesn't make that a bad app.
>
> Poor example because you're talking about a normal mode of operation.
> Put it in the DEP context where your 'Prime' program wants to fiddle
> directly with the TCP/IP stack and hack buffers... in a perfectly
> 'safe' manner of course. And then Mr. Hacker comes along and does the
> same thing except his 'purpose' isn't for 'speed or efficiency'.
So app B is bad, which still in no way makes app A bad.
>>>>That doesn't make such an application necessarily dangerous or bad.
>>>
>>>The application itself may not be malicious but if it depends on a
>>>methodology that necessitates bypassing a system's security features
>>>then it *is* 'dangerous and bad'.
No more so than when one occasionally has to tell NAV to allow a script
to run for an installation or poke a hole in a filrewall.
>> Not neccesarily. Is there and danger of Prime95 being exploited by
>> some third-party to execute data and comprimise the system? I've
>> seen no evidence of this.
>> Windows' DEP is a blanket general protection scheme
>> for to cover unknown bad apps. There are a lot of exploitable apps
>> out there--probably in no small part due to MS compilers. This scheme
>> was cooked up to cover those, but it gets some perfectly safe apps in
>> the process.
>
> You can attribute 'bad code' to anything you like but it's irrelevant.
> One can intentionally write 'bad code' as well and there's no way for
> DEP to know 'why' someone wrote code that's doing funky dunky things
> it shouldn't be doing; be it a compiler 'woops', a hacker, or the
> 'speed & efficiency' wizard.
But the user can know and in this case he would know that there is
nothing wrong with Prime95 and should be able to allow it to run.
>> One might argue that networking in general is dangeraous, as that's
>> the major path viruses and worms use to infect large numbers of
>> computers, but we don't disable all networking in XP, because that
>> would break a lot of useful and not dangerous apps and would be
>> unacceptable to most users. DEP, on the other hand, doesn't break
>> that much. If a few babies get thrown out with the bath water MS
>> doesn't really care.
>
> Inappropriate analogy. You compare removing entirely a vital
> capability, I.E. 'disable all networking', to simply having a program
> obey proper O.S. coding rules. There's nothing in DEP that prevents
> Prime from working, if it does so properly, and the proper analogy is
> that Prime obeying DEP 'rules' is equivalent to a firewall on that
> internet connection. It may not be quite as 'speedy or efficient', and
> it sure as heck isn't as convenient, but it's vital to system
> security.
As far as I can tell it obeys the rules and works properly. I made the
point that disabling all networking would be stupid and unacceptable. The
point of the analogy is that security features need to allow for
exceptions.
>> Back to your firewall example. If your contention is that the danger
>> comes in having to turn off DEP for all apps then my contention is
>> that DEP is broken. Just as your firewall would be broken if you had
>> to turn it completely off in order to surf the web.
>
> How is Windows to know, without so much as a hint, that 'Prime' is so
> privileged it's to be allowed O.S. level access? And how is it to know
> that this, lord knows where it came from, program is allowed but no
> others are?
We aren't talking about OS level access. The same way something like Zone
Alarm knows to allow your email client to run.
> The answer is: it can't. And if you allow Prime to wander unimpeded
> through system memory then you have to allow all the others too.
The DEP is bad, not Prime95.
> Again, your firewall analogy doesn't hold. A more appropriate analogy
> would be a firewall that allows just any old program to open any port
> it feels like any time it feels like it because the 'safe' program you
> write wants to do so for 'speed or efficiency'.
>
> I'm not concerned with the 'safe' program' It's the doors you want
> left open simply for that program's convenience. Of course, the next
> guy with a 'safe' program wants a different door left open. And the
> third guy...
Does a firewall allow all traffic to pass? No. It wouldn't be very useful
if it anly allow two options. Allow all traffic or none. That's why my
analogy is appropriate. Tat's the whole point.
>> I haven't touched the DEP settings in my PC and Prime95 works
>> perfectly well for me, however, so I'm not sure what the fuss is all
>> about.
>
> Probably something else entirely and I'm not saying DEP is the problem
> or that Prime is flawed. I'm just disagreeing with your assertion that
> DEP should allow just any old program that feels like it to romp
> through the system unimpeded. That defeats the whole concept.
Romping through the system unimpeded isn't what anyone is suggesting.
You're making a strawman argument. I don't want to allow just any old
program to become an open mail relay on my system, either, but I don't
argue that all TCP/IP functions are bad.