I am forwarding this to the CenPEG people, for their file.  Thanks Fooler.

//PManalastas


--- On Fri, 7/17/09, fooler mail <[email protected]> wrote:

> From: fooler mail <[email protected]>
> Subject: Re: [plug] Code Review & SysAdmin of Election 2010 Computers
> To: "Philippine Linux Users' Group (PLUG) Technical Discussion List" 
> <[email protected]>
> Date: Friday, July 17, 2009, 11:42 AM
> On Thu, Jul 16, 2009 at 11:26 PM,
> Pablo
> Manalastas<[email protected]>
> wrote:
> >
> > MY REACTIONS:
> >
> >
> > "After the customization of poll automation software"
> -- means the software already exists, so
> >why wait to finish customizing it (setting parameters)
> before getting the thing reviewed?  The
> >source code already exists, and will not be affected by
> customization, so why not review the
> >code now?
> 
> if their existing program is very flexible enough.. there
> is no so
> much customization in there.. the criteria is based on the
> parameters
> that is going to feed both for PCOS and CSS..
> 
> but comelec should start distributing the code of PCOS both
> the OS
> code and application code while waiting for CSS code..
> PCOS' OS must
> also be review especially on its TCP/IP stack for any
> backdoor access
> in case the machine is online from that start of voting..
> 
> 
> > If they are going to add new code, then that's a
> different story altogether. I'd like the following added to
> the code:
> >
> > 1) During initialization (hour zero on election day),
> i'd like both PCOS and CCS computers to
> >print out filenames of all executables
> >in /bin, /sbin, /usr/bin, /usr/sbin, /usr/local/bin,
> /usr/local/sbin, together with their SHA-256
> >checksums.  Also all configuration files in /etc and
> their SHA-256 checksums.  These are
> >needed to compare with the originals that have been
> approved by source code review.
> 
> for PCOS machine.. we suggest to them to simplify their
> partitioning
> scheme.. for example:
> 
> 1. /
> 2. /ER
> 3. /var
> 4. /tmp
> 
> all the binaries and configuration files must be in the
> root partition
> and that partition must be mounted as read only, executable
> and no
> writes.. in that way no updating of any files on that
> partition as
> well as we can easily make a hash starting from root as the
> message
> digest program will compute all the files in there
> recursively...
> 
> for /ER where the ER documents are stored in there.. this
> partition
> must be mounted as *append* mode only... append mode where
> you can add
> files and append data to an existing file but you cannot
> delete nor
> edit these existing files even if you are a root user...
> 
> for /var where the system and other logs are stored in
> there... this
> partition must be mount as *append* mode also just like the
> /ER
> partition...
> 
> for /tmp.. mounted as read/write but no execution of any
> files in there...
> 
> >In particular, /etc/securetty must specify that the
> root user must never be allowed to log in from
> >remote. To make checking simple for the BEI on election
> day, this printout must be in
> >lexicographic order of filenames, and must be digitally
> clear-signed by Smartmatic so that the
> >BEI need only compare the signature part (two lines of
> text)
> 
> so far i cannot comment or contribute as i need to know
> what is their
> plan about their network connectivity.. once i have a clear
> picture
> about on this... then thats the time for me to suggest...
> 
> >
> > 2) The option to SSL-sign or GPG-sign should be put in
> the code, so that the teachers have a
> >choice of getting their public keys signed by a CA or
> signed by a peer.
> > The signing part should be during the end of the
> voting period, not at the beginning when there
> >is nothing to sign yet (this is the way the current
> version of the program looks to me).
> >
> > 3) Support for Java smartcards (with CPU) in the PCOS
> and CCS hardware (smartcard slot)
> >and new program for signing, so that using Java
> smartcards for signing, the teachers' secret
> >keys never leave the card, and is never copied over to
> the computer.
> >
> > 4)  Others?
> 
> i took a look at the link of  technical details of
> PCOS machine gave
> by xander solis (thanks dude)... two graphics format are
> supported..
> bmp and tiff.. comelec is going to use tiff format..there
> is a well
> known vulnerability in tiff and can be seen from link below
> on exploit
> section...
> 
> http://en.wikipedia.org/wiki/Tagged_Image_File_Format
> 
> we need to code review not only the application program but
> as well as
> its API or libraries.. what the cheater need is to steal
> one private
> key and sign their special crafted ER where it can execute
> their
> arbitrary cheat codes secretly... in this regard.. a clean
> CSS code
> here is helpless for a honest election..
> 
> furtheremore, they need to replace their PCOS' gprs (2G)
> modem to
> HSDPA (3.5G) modem as this modem have a backward
> compatibility and
> fallback with gprs, edge and gsm networks...
> 
> please contribute guys if you have a better ideas aside
> from what is
> currently discuss here.. thanks..
> 
> fooler.
_________________________________________________
Philippine Linux Users' Group (PLUG) Mailing List
http://lists.linux.org.ph/mailman/listinfo/plug
Searchable Archives: http://archives.free.net.ph

Reply via email to