Re: Five Architectural Flaws in Windows Solved In Mac OS X



In article <0001HW.BF787F1000008691F0335550@xxxxxxxxxxxxxxxxxxxxxxxx>,
Tim Murray <no-spam@xxxxxxxxxxxx> wrote:

> On Sun, 16 Oct 2005 17:06:26 -0400, Daniel Johnson wrote
> (in article <2005101617062616807%danieljohnson@vzavenuenet>):
>
> >>> Sun can't implement that last (minor) feature because it would break
> >>> all the existing Unix programs that do not have code to "turn on" any
> >>> privileges.
> >>
> >> But they never have the security problems that windows has. Their RBACs
> >> work great. So does the old VMS Authorize utility to dole out
> >> priviledges to users. Windows still has break ins and is still the
> >> laughing stock of the industry with regards to security.
> >
> > Sure. Implementation counts more than design in this, and the exploits
> > that have been done on Windows have exploited flaws in the implementation,
> > not the design.
>
> But isn't implementation simply the follow-through of design? If a "designer"
> says "plug A into B" and it is implemented, then the issue is still design,
> right?

With real software, what happens is the designer usually writes a
high-level specification. For a network app, let's say, it could say
something like "Listen for incoming TCP/IP connections on port 1234. If
a client connects and sends the 8-bit ACSII string 'Yes', reply with the
8-bit ACSII string 'No'".

Now the implementor goes and picks a programing language and some
libraries, and decides how to store data read from the socket and how to
tell if that data is equivalent to the string 'Yes'. And maybe, for
instance, he reads data from that socket into a buffer in a language
like C, which doesn't do any bounds checking automatically, and he
doesn't bother to implement bounds checking himself. Oops. Now this
particular implementation of our little protocol has a buffer overflow
that could allow arbitrary code execution.

It wasn't the designer's responsibility to say 'And make sure to do
bounds checking!" The designer might not even know if that's a
meaningful concept in the programing language being used for
implementation.

--
"It's in our country's interests to find those who would do harm to us and get
them out of harm's way."
-- George W. Bush in Washington, D.C., April 28, 2005
.



Relevant Pages

  • Re: Five Architectural Flaws in Windows Solved In Mac OS X
    ... >>> privileges. ... >> But they never have the security problems that windows has. ... Implementation counts more than design in this, ...
    (comp.sys.mac.advocacy)
  • Re: M$ fixes 26 Security Holes
    ... agencies use Unix or OSX, they never consider Windows, since it's never ... that's why all security problems occur ... different design, one that doesn't allow for security issues Windows/DOS ... bar on January 19,2006 after losing his law license ...
    (comp.sys.mac.advocacy)
  • Re: M$ fixes 26 Security Holes
    ... since it's the basis of networking worldwide. ... Windows is fundamentally flawed, that's why all security problems occur ... different design, one that doesn't allow for security issues Windows/DOS ...
    (comp.sys.mac.advocacy)
  • Re: M$ fixes 26 Security Holes
    ... agencies use Unix or OSX, they never consider Windows, since it's never ... since it's the basis of networking worldwide. ... that's why all security problems occur ... different design, one that doesn't allow for security issues Windows/DOS ...
    (comp.sys.mac.advocacy)
  • Re: A 6% fix from Microsoft Security Bulletin MS03-040 - 828750
    ... >> the vendor knows of a defect that's being exploited In The Wild, ... That's why I see prudent modular design as being as much in MS's ... complete the job before Windows starts writing to disk. ... ignoring personal data scope; that's what SR tries to do. ...
    (microsoft.public.security)