flash_n_burn rom utils, /dev/bios ... resquest

Eric W. Biederman ebiederman at lnxi.com
Wed Apr 21 06:08:00 CEST 2004

Mathieu Deschamps <mdeschamps at mangrove-systems.com> writes:

> Le mer 21/04/2004 à 02:20, ron minnich a écrit :
> > On 20 Apr 2004, Ian Zimmerman wrote:
> > 
> > > Same here.
> > > 
> > > I feel that the reliance on hotplugging flash chips has limited
> > > linuxbios audience to hard-core metal hackers.
> > 
> > but what else can you do? Can you afford a flash burner? The set of people 
> > who can afford this is smaller than the set of people who can hot plug.
> 1. I can't afford a "true" flash burner
> 2. Hot plug on PLCC (for example) leaves me (and others as it seems)
> worry :
>      half a millimeter between chip pins and a working motherboard :
> what if i'am a bit stress and lift up badly the chip ? I got the
> answer...

In practice I have not seen this happen.  The only things I have
fried are BIOS chips themselves by pluggin them in backwards.
> It's where the difficulty relies : I find it a pity linuxbios offers so
> much and resolve via soft a lots of non Open/Free bios problems, and
> though couldn't free itself from this "mechanical" issue... if I can say
> so... :)

For users if there is a known good binary release for a board changing
BIOS chips is not an issue.  This is only an issue for developers, and users
willing to live on the edge because they don't know if what they are
about to flash will work.   

Once LinuxBIOS is up and running on a board we have two copies and you
can flash just so there is really not an issue.

> "Draft" Propositions :
> 1. why not building a flash_rom util which could send bios code steam
> via LTP, COM or USB, PCI  : the aim not to touch the "working" bios(even
> though its not needed after boot time) . Plusvalue of this : Cold plug :
> Make a backup copy for example

If this was done in software it would need to put the code some place.
That takes RAM.  You need the BIOS to enable RAM.  catch-22.  
> 2. trying to disconnect/diswire bios socket from the up mainboard (that
> is to say to get 0v on every pin of the socket) . Ideal view (mb
> manufacturers !) : get a jumper to handle this 

Sounds like a sane alternative to me, if you can solder that well.

> 3 ... err .. run dry

Right. We try to provide suggestions within the range of doability.
It is not that we want to make things hard on developers/testers it
is just that things are hard and tricky taking over a system that
low down.


More information about the coreboot mailing list