On Mon, 9 Jul 2001, Ronald G Minnich wrote:
> On Mon, 9 Jul 2001, Armin Schindler wrote:
>
> > When I use the standard BIOS which comes with the evaluation board,
> > the 'hlt' check is no problem.
>
> It's probably an IRQ problem then. I am pretty sure that the 'hlt' check
> counts on the clock interrupt to get it out of the halt. Kernel does a
> 'hlt' with interrupts enabled, clock interrupt happens, kernel wakes
> itself up. Not clock interrupts ==> no wake up.
Yes, I think so too.
> > Do I have to activate/configure interrupts(controller) stuff
> > before the kernel can be started ?
>
> you definitely have to make timer interrupts work right. For 2.2 you don't
> have to do much. For 2.4 you only need to do one or two things, but clock
> interrupts MUST be done right.
But isn't that flawed, that I (the BIOS) have to do different
init for different kernels ? It seems to me a kernel bug, when
the kernel checks for something, which is done later anyway.
> > What I don't understand is, that after the 'hlt' check
> > (when check is disabled), everything else works. It seems
> > that the kernel expects something set when hlt is checked, and
> > afterwards the kernel sets it by itself !?
>
> which kernel version. This is a good description of the kind of things
> that happen now. 2.4.x does a lot of work now, but early kernel code still
> relies on certain things having been set up, and timer interrupts have
> always been a problem.
I use 2.4.3. Does anyone know on what this "early code" relies on
exactly ?
Armin
