[Cryptech Tech] arm

[Basil Dolmatov]

dol@ с iPad

> 18 янв. 2015 г., в 11:40, Peter Gutmann <pgut001 at cs.auckland.ac.nz> написал(а):
> 
> Basil Dolmatov <dol at reedcat.net> writes:
> 
>> Following some set of design rules and protocols on the bus will make this
>> vector of attack us feasible.
> 
> (I assume "us" should be "less").
Yes, blame the autocorrection...
> 
> It doesn't work that way.  Look at, for example, this diagram:
> 
> http://m.eet.com/media/1092513/ATMEL_SAM3U_Block_Diagram.jpg
> 
> The "USB device" IP block in there will be (say) 8051 with its own ROM and RAM
> implementing the USB protocol.  If you can compromise that, you have DMA
> access to everything in the CPU/SoC.  With a USB/SPI bridge you'd need to
> compromise the USB bridge, use that to compromise the SPI bus, and then get
> across the peripheral bridge before you can get the same level of access (if
> that level of access is even possible) that the on-chip USB gives you.
That's "design rules"... 
If peripheral can overtake the bus and influence CPU, this means  that this given design sucks (from the security point of view, it can be very cool from speed or convenience point of view). 
> 
> Peter.