patch

[joel at oarcorp.com (joel at oarcorp.com)]

On Tue, 1 Sep 1998, Eric Valette wrote:
> 
> Gentlemen, I'm affraid I will play again the role of an old grandfather
> but there is an IEEE standart for BOOT (initialization Configuration)
> firmware IEEE 1275-1994.

Not that old. :)

Isn't this the Open Firmware standard?

> This firmware is used on nearly all modern PPC board, Sun sparc
> manufactured
> board, ... This standarts defines many thing you can do in order to
> setup a 
> running system and also how to retrieve the configuration variable from
> the
> NVRAM.
> 
> In particular it defines how to chose :
> 
> 	- the boot devices,
> 	- the boot file,
> 	- input device,
> 	- output device

The key to the libchip probe stuff is that if you have NVRAM configuration
you can probe using standards based routines, fabulous.  We may end up
with a standard set of probe routines.  On other boards, these may do
nothing more than read a board register which reflects jumpering.  

> I agree with alexsey that you may want to choose the output device 
> independently of the hardware capabilities. I disagree with him that
> printk should have an different output than printf...

Chris Johns has a lot of experience with logging in complex environments.
Hopefully he will wade in on this one. :)

> As far as I'm concerned, I appreciate his patch. I just included a
> very small patch that needs to be applied on top of its own patch to
> fix the following things :
> 	- IBMPC_Init_video was called twice,
> 	- a cosmetic change to make the cpu.c code more easy to understand
> 	limit = (limit + 1) >> 3; => limit = (limit + 1)
> /sizeof(interrupt_gate_descriptor);
> 
> The funny things is that the gererated code is the same as ggc is smart
> enough to use shift to divise by 8...

gcc is smart and getting smarter. :)

--joel