Re: A20 delay

[Quality Quorum <qqi@world.std.com>]

On Thu, 16 Dec 1999, VALETTE Eric wrote:

> >>>>> "Aleksey" == Quality Quorum <qqi@world.std.com> writes:
> 
> Aleksey> As far as I understand previous patch I submitted was corrupted
> Aleksey> when merged with recent snapshot. Are you referring to something else ?
> 
> Yes. We fixed the not closed comments and tried your delay with the loop
> implementation. We get an error when programming the 8259 leading
> to incorrect interrupt controller behavior and a system with no interrupt 
> at all.

????????? Wow. 


> 
> If instead of the "delay loop" we use the old "outb al, $0x80" delay, it works again...
> 
> As far as I understood, the delay was causing problem to you for A20 setting and
> not the PIC. If this is the case, I would suggest to let the delay as 
> "outb	al, $0x80" by default in ldseg.S. Whatever you do in start16,
> I do not care : I do not use it. BUT I would suggest to look at linux 
> code as an example of *highly tested* solution.

It is fine with me, however, I feel somehwat uneasy that having delays of 
say 10us vs 1us will cause 8259 programming errors - it may be a sign that
this 8259 delay sensitivity is worth looking at. I am wondering is 
it reproduceable across other systems ? What chipset are you using it 
with ?

>   /  `                   	Eric Valette

Thanks,

Aleksey