Re: Windows 2000 Source Leak Verified. Get ready for the havoc.

[Nick FitzGerald <nick () virus-l demon co uk>]

Valdis.Kletnieks () vt edu wrote:

> On Sat, 14 Feb 2004 16:42:39 GMT, Lee <cheekypeople () sec33 com>  said:
> > again its 1/100 of standardd MS code for a OS, lets get a grip please... and
> > I think I see the company who let the source get loose come out and say
>
> Most earlier estimates of the Win2K source were about 45M lines of code (I think
> the "40 gig" being tossed around is the size of the source-control-system database).
> And I've seen the number 12.5M lines of code escaped.  That's closer to 1/3 than
> to 1/100.

Indeed -- there are some public domain references (from folk who should 
know) about the size of the source control system, but of course that 
is way more than just the final source.  Using the 45M lines of code 
estimate (was it reallY??  perhaps for server with IIS, etc?), some 
simple maths tells us something about the expected size of the source.

Assume an average of 60 characters per line (reasonable??), assume a 
simple .ZIP compression ratio of 70% (seems a tad low for .C source 
based on some large .ZIPs I just checked on this machine -- most ran
72-75%):

  45M * 60 * 0.30 = 810MB

Vary to suit your tastes regarding likely average line length and .ZIP 
compression ratio...

Also, consider that the directory listings I've seen posted for the 
"leaked" .ZIPs show that there is quite a bit of cruft included (.EML 
files and other non-source stuff like core dumps) and make further 
adjustments.  So, it seems that the "leaked" source is considerably 
more than 1/100th the original.  I think Valdis' 1/3 estimate may be a 
tad low as I think I saw the 12.5M lines estimate for the NT code base. 
Ahh, yes -- Russ Cooper posted the following to NTBugtraq:

   1. NT source is NT 4.0 SP3, contains 27000+ files (658MB). It is all
   NT 4.0 Server except IIS, includes IE 4. No references to Mainsoft
   (see http://www.eweek.com/article2/0,4149,1526830,00.asp.)

   2. W2K is SP1, a very small subset, IE 5, SNMP, PKI, networking and
   some SDK stuff. 28000+ files (338MB - although many of these are
   empty mail messages and other crap.) Does contain 3 references to
   MainSoft. Much of what is there is available elsewhere.

and in another message Russ wrote:

   Couple of corrections.

   1. There were 27,142 NT 4.0 SP3 files totaling 338MB.
   2. There were 28,782 W2K SP1 files totaling 658MB.
   3. It does appear that all of both versions are present, minus IIS.
   4. 10,425 of the 27k NT files are actually source totaling 193MB
      uncompressed.
   5. 8,367 of the 28k W2K files are actually source totaling 217MB
      uncompressed.

Archived copies of the full messages from which these comments were 
extracted are at (sorry, URLs will wrap):

http://www.ntbugtraq.com/default.asp?pid=36&sid=1&A2=ind0402&L=ntbugtraq
& F=P&S=&P=2868  

http://www.ntbugtraq.com/default.asp?pid=36&sid=1&A2=ind0402&L=ntbugtraq
&F=P&S=&P=2954


A followup comment by Dragos Ruiu may be of interest too:

http://www.ntbugtraq.com/default.asp?pid=36&sid=1&A2=ind0402&L=ntbugtraq
&F=P&S=&P=3155


Regards,

Nick FitzGerald

_______________________________________________
Full-Disclosure - We believe in it.
Charter: http://lists.netsys.com/full-disclosure-charter.html