[GLLUG] Am I over-reacting to this?

[Matthew Walster]

On 16 January 2014 10:05, James Courtier-Dutton <james.dutton at gmail.com>wrote:

> I think UEFI bios has made things worse in this area. UEFI is so large
> and difficult to decompile.
> I think the BIOS code would be much simpler if it was a small as
> possible, and the user would install into flash their boot image. I.e.
> vmlinuz and initrd.img
>

​That's essentially what UEFI is -- it gets the system from an empty shell
to a point where the hardware is in a state to be used, then it reads a
series of instructions from the primary boot device as to what to do next.
There's no point in "flashing" it to an EEPROM as you want to update it
fairly regularly, so there's usually a reserved "EFIBOOT" area or similar
with a simple set of instructions that tell the system what to do next.

This is essentially the same as a modern "BIOS", and GRUB stages 1/1.5/2
all wrapped into one, with some additional functionality built in too.

UEFI isn't bad by itself, but as you rightly say it has been very poorly
implemented by a number of companies.

There's a very interesting talk by Matthew Garrett on this:
http://www.youtube.com/watch?v=V2aq5M3Q76U​


> The BIOS would then only need the code to switch on the CPU, have a
> hardware description table, and then load the next boot stage from
> flash.
>

​That's exactly what a BIOS does today -- except it still has to expose all
(ok, almost all) the stuff it had to do back in the 8086 era, interrupts
and all: http://en.wikipedia.org/wiki/BIOS_interrupt_call​



> What I would like to see is more openness from hardware manufacturer.
> E.g. A motherboard manufacturer should publish a standards based
> computer readable description of their motherboard.  I.e. Which CPU,
> which interface chips, over-clocking registers etc.
>

​It already exists. It's called a BIOS, along with the information that you
can pull out using "dmidecode" that are stored in other registers.


> The open source community could then write their own BIOS that used
> the manufacturers "description" to compile a BIOS that was certain to
> boot that motherboard.
>

I know http://en.wikipedia.org/wiki/OpenBIOS tried to do some work on this,
I'm not sure how far they got.

An interesting topic, nonetheless.

M​​
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.lug.org.uk/pipermail/gllug/attachments/20140116/f348040c/attachment.html>