[Arm-netbook] I Have An Possible Chance To Meet An ARM Senior Manager

[Gordan Bobic]

On 10/08/2012 12:02 PM, luke.leighton wrote:
> On Mon, Oct 8, 2012 at 1:47 AM, Alexander Stephen Thomas Ross
> <arm-netbook at aross.me>  wrote:
>
>> Hang on ARM stuff does not have a BOIS. They don't need one. They have a
>> boot thingly like uboot which I believe is (100%) free software.
>
>   yeah - gordan put it well, by pointing out that u-boot is actually a
> modified copy of the linux kernel source code: it has to be, otherwise
> you can't boot up and load from the NAND flash, because you don't know
> how to access it, right?
>
>   why in god's name u-boot was even written, rather than just having an
> initrd with an application that presents a simple menu and calls kexec
> when the user selects what to do, i really don't know.

Precisely, it's all a bit... "special"... for want of better word.

>   if you've seen the marvel-based "openrd" nightmare situation where
> you can boot from NAND but not from SD, yet after booting you *can*
> access the SD, you'll know that this is down to lack of proper
> maintenance of u-boot.

It's not the maintenance as much as multiple custom forks of uboot that 
are the problem. There is no single unified uboot that will run on all 
uboot devices from what I can tell.

> whilst the drivers in linux get SD working,
> they haven't been properly cut over to the
> version-of-the-exact-same-drivers-running-in-uboot.
>
>   so actually, u-boot makes the situation *worse* not better.

I'm not sure if it's _worse_ than having an unbootable device, but it 
makes a mockery of any notion of standardisation since no two uboot 
devices are the same in terms of boot media and access to it, console, 
or memory mapping. It's not a standard at all - it's a horrible bodge of 
a _lack_ of standards.

Gordan