Re: [nmap-svn] r19501 - nmap-exp/luis/nping-echo

["Luis MartinGarcia." <luis.mgarc () gmail com>]

On 08/06/2010 02:56 AM, David Fifield wrote:
> On Thu, Aug 05, 2010 at 05:51:18PM -0700, commit-mailer () insecure org wrote:
>   
> > Author: luis
> > Date: Thu Aug  5 17:51:18 2010
> > New Revision: 19501
> >
> > Log:
> > Fixed a problem that caused Nping to abort on Windows due to a stack
> > overflow. After hours looking for the problem, I found that it was
> > Visual Studio's fault because by default it only allocates 1MB stacks.
> > Incrementing this value to 5MB (quite reasonable for today's RAMs)
> > does the trick.
> >
> > Modified:
> >    nmap-exp/luis/nping-echo/nping.vcproj
> >
> > Modified: nmap-exp/luis/nping-echo/nping.vcproj
> > ==============================================================================
> > --- nmap-exp/luis/nping-echo/nping.vcproj    (original)
> > +++ nmap-exp/luis/nping-echo/nping.vcproj    Thu Aug  5 17:51:18 2010
> > @@ -82,6 +82,7 @@
> >                              GenerateDebugInformation="true"
> >                              ProgramDatabaseFile=".\Debug/nmap.pdb"
> >                              SubSystem="1"
> > +                            StackReserveSize="5000000"
> >                              OptimizeForWindows98="0"
> >                              RandomizedBaseAddress="1"
> >                              DataExecutionPrevention="0"
> >     
> Interesting. Do you have any idea how the stack gets that deep in the
> first place?
>
> David Fifield
>   

Well, I honestly don't know. Nping doesn't have any significant memory
needs. Probably the biggest structure out there is the EchoHeader class,
which has a big internal buffer (I'll try to reduce that). However, I've
been doing some tests on linux and Nping consumes a lot less memory than
Nmap (obvious, since it does not have to keep as much state as nmap), so
I don't really know why I had to increase its stack size on Windows. 
These are the results of my tests:


"TOP" COMMAND STATISTICS

NPING (nping --echo-client mypass echo.nmap.org -d4 --delay 500s)
  PID  VIRT  RES  SHR S %MEM P SWAP CODE DATA nFLT nDRT
13351 39740 8324 6332 S  0.2 0  30m  372 2888    0    0


NMAP (nmap scanme.insecure.org -p80)
  PID  VIRT  RES  SHR S %MEM P SWAP CODE DATA nFLT nDRT
13360 50104  14m 2936 S  0.4 0  34m  996  12m   12    0


PID        = Process Id
VIRT       = Virtual Image (kb)
RES        = Resident size (kb)
SHR        = Shared Mem size (kb)
%MEM       = Memory usage (RES)
P          = Last used cpu (SMP)
SWAP       = Swapped size (kb)
CODE       = Code size (kb)
DATA       = Data+Stack size (kb)
nFLT       = Page Fault count
nDRT       = Dirty Pages count



VALGRIND STATISTICS

NPING (nping --echo-client mypass echo.nmap.org -d4 --delay 500s)
==8935== HEAP SUMMARY:
==8935==     in use at exit: 27,826 bytes in 50 blocks
==8935==   total heap usage: 5,207 allocs, 5,157 frees, 820,636 bytes
allocated


NMAP (nmap scanme.insecure.org -p80)
==13372== HEAP SUMMARY:
==13372==     in use at exit: 1,178,248 bytes in 37 blocks
==13372==   total heap usage: 60,339 allocs, 60,302 frees, 5,215,047
bytes allocated


Regards,

Luis MartinGarcia.


_______________________________________________
Sent through the nmap-dev mailing list
http://cgi.insecure.org/mailman/listinfo/nmap-dev
Archived at http://seclists.org/nmap-dev/