Slimelite BBS

bios wiping?!?!

Back to C_PLUSPLUS

From Jeffrey Bloss · 1996-04-07 22:28 · #75079 in C_PLUSPLUS

DM>>  JB> All the non-jumpered flash BIOS implementations I've come
DM>>  JB> across or heard about require three things to start the flash
DM>>  JB> upgrade process... a cold boot, a keypress combination, and a
DM>>  JB> special floppy disk.

DM>> Seems to be a software protection...

Firmware... with what amounts to a hardware switch implemented via the
keyboard because of the necessity of a POST error.  As far as I know
there's no way to stuff the keyboard buffer, then cold reboot the
machine and still have the buffer holding the keystrokes. ;)

DM>> And what if the Virus simply implement the flash process routine as it is
DM>> stored in the Bios/floppy ???

If I understand what you're getting at, we'd be talking about an
infected flash upgrade.  Very different animal than a virus that could
access flash BIOS all by itself.

DM>> Or directly accessing the ports via IN/OUT commands to enable the BIOS
DM>> writing

They're not port reads and writes... more of a "DMA thingy", and the
entire process is accessible to existing BIOS code only after the BIOS
has been "unlocked" by a series of events that are triggered by a
specific hardware failure condition.

DM>> The only "software controlled" mean to protect the bios is the following :
DM>> have a flip-flop that is resetted when the RESET button is pushed (it's

I would think depending on the RESET switch is asking for trouble... you
can simulate this sorta reboot in software.  Every non-jumper flash BIOS
upgrade procedure I've seen requires a complete power down.

I'm not saying the possibility of a flash BIOS virus is out of the
question, just that from what information I can gather, and from some
amount of close examination of a couple flash implementations, I haven't
found one that I think could be infected by a standalone piece of
software yet. Of course this doesn't take into account the possibility
of a flash BIOS "back door" either.  A flash-critter may in fact exist,
and to be sure there's code out there that *claims* to be able to infect
a flash BIOS... but the 4 contenders *I've* come across simply do not
work.

                   Jeffrey Bloss -=- jbloss@meadville.com
                   PGP: 8DF9C2C36732A9A0-02C216F823D52D42

--- WILDMAIL!/WC v4.00 
 * Origin: Meadville Online  (1:2601/551.0)