Am 06.06.2011 20:14, schrieb Darren New:

> For that matter, there's very few systems out there that require all
> your code to be digitally signed and which checks it every time you fire
> it up, other than game consoles and phones for copy-protection purposes.

... and still, even /those/ seem to keep dropping their pants if 
addressed in the right manner. And you'd think, if anyone is willing to 
invest real money into such systems, it would be the makers of such 
systems (apart from military or secret agencies of course).

Seems like the problem is not /that/ easily solved...

Post a reply to this message

On 6/6/2011 11:10, Orchid XP v8 wrote:
> Actually, last time I tried it, you can't do this in Windows either.

Not only can you dismount it, when you remount it, you have the same cwd.

Since every drive has a cwd, it would be impossible to dismount any drive if 
you couldn't dismount one where you had a cwd set.

>> How come every time you come across legacy support, you say it's a
>> kludge or a wtf or a random? :-)
>
> What was a good idea 40 years ago is not necessarily a good idea today.

That doesn't make it a kludge or a wtf or a random. It makes it an idea 
whose time has passed.

-- 
Darren New, San Diego CA, USA (PST)
   "Coding without comments is like
    driving without turn signals."

Post a reply to this message

On 6/6/2011 12:51, clipka wrote:
> ... and still, even /those/ seem to keep dropping their pants if addressed
> in the right manner.

Usually through the use of a hardware hack, tho.

The Sony hack required him to actually wire up the motherboard with  switch 
shorting out traces. The XBox hack was, IIRC, not so much a flaw in the 
security of the system, but a broken game that let you run code from a saved 
game (via buffer overflow or some such).

That's exactly why you need to do something like formal logical checks that 
your code does the right thing at the lowest levels, then make everyone 
conform to that.

When you're talking security against malware rather than copy protection, 
it's a lot easier to get right, in some ways, because you don't *ever* want 
the malware to run.  Nobody is going to be sticking debuggers on the chip or 
shorting out motherboard traces in order to get malware running on their own 
machine.

-- 
Darren New, San Diego CA, USA (PST)
   "Coding without comments is like
    driving without turn signals."

Post a reply to this message

Am 06.06.2011 22:15, schrieb Darren New:
> On 6/6/2011 12:51, clipka wrote:
>> ... and still, even /those/ seem to keep dropping their pants if
>> addressed
>> in the right manner.
>
> Usually through the use of a hardware hack, tho.
>
> The Sony hack required him to actually wire up the motherboard with
> switch shorting out traces. The XBox hack was, IIRC, not so much a flaw
> in the security of the system, but a broken game that let you run code
> from a saved game (via buffer overflow or some such).
>
> That's exactly why you need to do something like formal logical checks
> that your code does the right thing at the lowest levels, then make
> everyone conform to that.

Ah yes - a formal proof... pretty useful if your intention is to make 
sure a security-critical system never fails due to unforeseen errors.

As for making sure that a system is secure against /malicious intent/, I 
believe it's pretty useless.

Just have a look at smart cards: For a given smart card design, you may 
formally prove that there is no input sequence that makes the card 
disclose even a single bit of its secret key...

... on its /official/ interface, i.e. the I2C bus data lines. But such a 
formal analysis typically forgets a few other channels on which the 
device is leaking information. For instance, an analysis may forget 
about the timing of the data output, which might give hints about what's 
going on inside. Or the power the chip consumes at any given time.

Say you will - I think a formal analysis can never foresee /all/ 
possible attack vectors a system might exhibit.

Post a reply to this message

On 6/6/2011 15:13, clipka wrote:
> Ah yes - a formal proof... pretty useful if your intention is to make sure a
> security-critical system never fails due to unforeseen errors.

That is the intention.

> As for making sure that a system is secure against /malicious intent/, I
> believe it's pretty useless.

No, it's very good at that.

> ... on its /official/ interface, i.e. the I2C bus data lines.

Naturally. All you're saying is that a formal system won't break as long as 
the underlying system matches the formalism.

It's *much* more difficult to mount an attack against a smart card when 
you're not physically in possession of the smart card than when you are. 
Plus, again, the attacks on the smart card aren't the same category as 
malware. With DRM, the card (or console or whatever) has to be able to 
perform its operations while hiding its results from observers. With 
malware, the system doesn't want to perform the operations at all.

The system relies on programs to obey the rules that the system enforces, 
just like your operating system relies on the hardware doing what the 
hardware says it does.

> Say you will - I think a formal analysis can never foresee /all/ possible
> attack vectors a system might exhibit.

It depends on the kind of attack you're trying to prevent. It's easy to show 
that formal analysis can foresee you never running off the end of an array 
to access memory belonging to another process. How much malware have you 
seen that takes advantages of flaws in the CPU mask? Formal analysis can 
foresee all possible attack vectors for particular attacks, assuming the 
mathematical system is isomorphic with the physical system. Of course if 
you're trying to hide information in one program from being observed by 
another, you have to take care that things not accounted for in the 
formalism (such as the timing) don't happen. But the things accounted for by 
the formalism can be shown not to happen, as long as the hardware (et al) 
obeys the same rules as the formalism.

In other words, it becomes far easier to check you're right, because you've 
written a very few number of rules that you have to manually check are 
correct, and the computer deduces from them that the top-level properties 
you want to hold do indeed hold.

And there have been formally-verified chip designs as well, where the 
assembly language was expressed in math and they proved that the chip mask 
made the chip follow the spec. (To some extent, I assume. I would imagine it 
would be difficult to prove that, for example, what you etched is what you 
thought you etched.)

-- 
Darren New, San Diego CA, USA (PST)
   "Coding without comments is like
    driving without turn signals."

Post a reply to this message

On 06/06/2011 08:52 PM, Darren New wrote:
> On 6/6/2011 11:10, Orchid XP v8 wrote:
>> Actually, last time I tried it, you can't do this in Windows either.
>
> Not only can you dismount it, when you remount it, you have the same cwd.
>
> Since every drive has a cwd, it would be impossible to dismount any
> drive if you couldn't dismount one where you had a cwd set.

Oh I see. You're saying that you can dismount it even if there's a 
remembered path. I'm saying you can't dismount it if it's the *current* 
path.

>> What was a good idea 40 years ago is not necessarily a good idea today.
>
> That doesn't make it a kludge or a wtf or a random. It makes it an idea
> whose time has passed.

A kludge to rush something to market a few months earlier is still a 
kludge 40 years later, yes.

Post a reply to this message

On 6/7/2011 1:09, Invisible wrote:
> Oh I see. You're saying that you can dismount it even if there's a
> remembered path. I'm saying you can't dismount it if it's the *current* path.

Oh, I see. This changed over time to actually include a lock on the current 
path. It didn't always used to do that.

> A kludge to rush something to market a few months earlier is still a kludge
> 40 years later, yes.

Well, in that event, pretty much everything in the world is a kludge, 
including every human being (who is born some 9 months earlier than they 
should be so the head can fit through the birth canal.)

-- 
Darren New, San Diego CA, USA (PST)
   "Coding without comments is like
    driving without turn signals."

Post a reply to this message

>> A kludge to rush something to market a few months earlier is still a kludge
>> 40 years later, yes.
>
> Well, in that event, pretty much everything in the world is a kludge,
> including every human being

Oh hey, biology is, like, the *ultimate* super-kludge!

Then again, biology wasn't designed by a concious intelligence.

Post a reply to this message

Now here's another random thing:

The syntax for a drive is a letter followed by a colon.
The syntax for a device is a few letters followed by a colon.
The syntax for an ADS is a file name followed by a colon followed by a 
stream name.

So consider the string "C:Foo".

Is that the file "Foo" in the current directory on "C:"?
Or is that a file named "C" with an ADS named "Foo"?

Answers on a postcard...

Post a reply to this message

On Wed, 08 Jun 2011 11:57:48 +0100, Invisible wrote:

> So consider the string "C:Foo".
> 
> Is that the file "Foo" in the current directory on "C:"? Or is that a
> file named "C" with an ADS named "Foo"?

It's just a string.  The context in which the string is used is important.

If "Foo" is a file, then it would hardly be stored as the CWD in memory 
on the machine - for example.

Jim

Post a reply to this message