Re: Modifying GRUB to read DMI data

[W. Michael Petullo]

On Aug 24, 2008, at 6:32 AM, Javier Martín wrote:

> 2008/8/24 W. Michael Petullo <address@hidden>:
> > > > I am interested in modifying GRUB to read DMI data from my BIOS
> > > > (i.e., the data dmidecode reads). The SMBIOS specification seems to
> > > > state that the Plug-n-Play function interface is deprecated  
> > > > while the
> > > > table-based interface requires a processor running in 32- or 64-bit
> > > > protected mode. Does anyone know where I could find information  
> > > > on a
> > > > non-deprecated method to read DMI data while in real mode?
> > >
> > > But you just said this can be done by parsing tables in i386 32- 
> > > bit mode,
> > > why not just do that?
> > >
> > > Btw, which information do you need to obtain from there?

> > I was under the impression that the DMI tables could be stored  
> > above 1MB,
> > and that these memory locations were not easily accessible when in  
> > real
> > mode.
> >
> > I have now found that, at least in my hardware's case, the DMI  
> > tables are
> > below 1MB and easily read. Is this the case for all hardware?
> >
> > I have a client that has a piece of hardware that provides an  
> > electronic key
> > via DMI when a physical key is inserted into the device. I am  
> > writing a
> > command that will read that key and store it in GRUB's environment  
> > for
> > Michael Gorven's LUKS module to pick up.

> The environment of a GRUB module is 32-bit pmode - you can "directly"
> access DMI: why do you need real mode?

I don't need real mode. But, I thought GRUB ran in real mode and  
protected mode was not enabled until the kernel enabled it. So, it  
looks like this is much more straightforward that I had thought.

Mike