Re: [Mspgcc-users] JTAG Security Fuse

[Travis G. <tr...@ut...>]

Howdy Gunther,

You're right that BSLKEY moves around.  It resides at 0xFFBE on the 4618
with BSL 2.12, 0xFFDE on the 2274 with 2.01. What variety!

Early in the BSL, there's a comparison of that value to 0xAA55, followed
by an infinite conditional jump.  Thus when BSLKEY=0xAA55, the BSL hangs
in an infinite loop.
cmp     #0xaa55,&0xffbe
jz      $+0

"A zero disables the erasure of the flash if an invalid password is
supplied."
That is a horridly confusing way of saying that the Mass Erase command
usually does not require a password.  By setting it to zero, you can
password protect this function.  I'd rewrite the statement as
"A zero disables the Mass Erase command until a valid password has been
supplied."

At least as of 2.12, there's no booby-trap that erases memory after too
many incorrect passwords.  The "brute force cracking protection" is just
a fix of the Change Baud Rate design mistake of 1.60 and 1.61.

Cheers,
--Travis Goodspeed

Gunther Lemm wrote:
> Hi Travis,
>
> thanks for the detailed info on the BSL key. For all the other guys here 
> who didn't follow your link to SLAA089:
>
> "The password itself consists of the 16 interrupt vectors located at 
> addresses FFE0h to FFFFh (256 bits), starting with the first byte at 
> address FFE0h. After mass erase and with unprogrammed devices, all
> password bits are logical high (1)."
>
> The note in the MSP430F248 datasheet (below the interrupt vector table 
> assignment) is somewhat misleading:
>
> "The address 0xFFDE is used as bootstrap loader security key (BSLSKEY).
> A 0xAA55 at this location disables the BSL completely.
> A zero disables the erasure of the flash if an invalid password is 
> supplied."
>
> This seems to be the register that's not always located at the same 
> address. MSP430FG461x devices have their DMA interrupt vector at 0xFFDE.
>
> Do you know anything about that flash erasure on invalid passwords? I 
> guess not, cause otherwise you would've seen it while disassembling the 
> BSL code. So what does TI mean with that note?
>
> Gunther
>
> Travis Goodspeed schrieb:
>   
>> Howdy y'all,
>>
>> I've been doing a bit of work with the BSL, mostly Version 2.12 which I
>> dumped from an MSP430FG4618, disassembled, and annotated.
>>
>> --The BSL password isn't beneath the IVT, it *is* the IVT.
>>
>> --There is an unprotected command, Mass Erase, which erases all of
>> memory.  The idea is that you should be able to replace the firmware,
>> but not to extract the firmware, as you could always replace the whole
>> chip on any board.  Once memory is erased, every bit of the IVT--and
>> thus the password--becomes 1.
>>
>> --In very recent versions of the BSL, you can set a flag in flash to
>> password-protect the Mass Erase command.
>>
>> --Versions 1.60 and 1.61 are the best candidates for brute-forcing, as
>> they have an unprotected command, Change Baud Rate, for writing directly
>> to the clock registers.  Earlier versions have no such command, while
>> later versions require the password to be sent before the baud rate is
>> changed.  Supposing only 40 bits of the password are random and you are
>> attacking a chip which is clocked to 16mhz, it will still take 32 years
>> to guarantee a break.  It can be made a bit faster, but not so much as
>> to make brute forcing practical.
>>
>> Gotchas:
>> --Version 2.12 is vulnerable to a side-channel timing attack.  2.01 and
>> earlier are not.
>> --If you are blowing JTAG and expect the BSL to protect cryptographic
>> keys, realize that the key is not part of the IVT and therefore it is
>> not part of the password.
>>
>> --rand_int.pl, attached, will randomize the interrupts by making them
>> point to branch instructions that direct back to the real address.
>>
>> For more details BSL brute forcing, see
>> http://travisgoodspeed.blogspot.com/2008/06/msp430-bsl-passwords-brute-force.html
>> For the official docco,
>> http://www.google.com/search?q=slaa089
>>
>> Cheers,
>> --Travis Goodspeed
>>
>> Gunther Lemm wrote:
>>     
>>> There was a discussion about the security fuses and some enhancements of 
>>> the BSL password protection on the MSP430 day 2008 in berlin. I only 
>>> remember some basic infos:
>>>
>>> - you can only blow the JTAG security fuse once because it's a physical 
>>> fuse. so there's no way to reset these fuses.
>>>
>>> - the BSL password is stored at a memory address just before the 
>>> interrupt vector table (take a look at the device specific datasheets 
>>> because the location depends on 32/64 byte vector table size)
>>>
>>> - AFAIR there is some brute force password cracking protection in 
>>> msp430f2xxx devices
>>>
>>> - there is a way to reset the BSL password, but this also deletes the 
>>> complete flash contents (don't ask me how to do it)
>>>
>>>   
>>>       
>> ------------------------------------------------------------------------
>>
>> -------------------------------------------------------------------------
>> Check out the new SourceForge.net Marketplace.
>> It's the best place to buy or sell services for
>> just about anything Open Source.
>> http://sourceforge.net/services/buy/index.php
>>
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> Mspgcc-users mailing list
>> Msp...@li...
>> https://lists.sourceforge.net/lists/listinfo/mspgcc-users
>>     
>
>
>