Ronald G Minnich <[EMAIL PROTECTED]> writes:
> On Mon, 27 Aug 2001, Ollie Lho wrote:
>
> > In LinuxBIOS we init the i8259 with i8259.inc. But in the kernel
> > the same code is executed again. Can the init code been run twice
> > without hardware reset for i8259 ?? Does i8259 consider the 2nd time
> > init as some knid of "operation programming" (probably illeagle) ??
Nope. At least that is the experimental evidence I have.
mkelfImage includes this code so it can boot 2.2 and I don't
have problems booting 2.4 which does this in C. I haven't been using
linuxBIOS but I have been using mkelfImage in production all year
long.
I'd have to double check but I believe what we do is a software
reset and then program the interrupt controllers, to default values.
Also it would be a bug for the kernel to unconditionally do a one
time only hardware operation. And we stole this code from the kernel.
> well, good question. But without that init in linuxbios we never get
> through the bogomips test, because there are no clock interrupts.
This is problem is specific to 2.2. Ron this is something that needs
to be double checked with my code rearrangements. In particular I
moved this code to C and disabled running it by default. We have
never strongly supported 2.2 which is why it hasn't shown up until
now.
But this is something that should not just be forgotten as I did. :)
Eric
