Re: [PATCH] Automatic RAM detection for VT8601

Hi !!!!

I have a question about ADLO. My LinuxBIOS + etherboot + ADLO go through
to ADLO, but stop at " sti " instruction  before jumping to Bochs BIOS.
I look up "sti" from Intel IA32 Instruction set Volume 2.  I read that
"sti" is "protected mode virtual interrupt enable".

I removed "sti" from ADLO because it is real mode. 
Does It affect badly to my EPIA operation ?

ps.> If I reset my system with LinuxBIOS after booting it with PC BIOS,
it works great.
As result , What makes it works ?

Need your help... Thanks.

regards
newbie

[PATCH] Automatic RAM detection for VT8601

[SONE Takeshi]

On Tue, May 20, 2003 at 02:15:23PM +0900, [e-mail] wrote:
> I have a question about ADLO. My LinuxBIOS + etherboot + ADLO go through
> to ADLO, but stop at " sti " instruction  before jumping to Bochs BIOS.
> I look up "sti" from Intel IA32 Instruction set Volume 2.  I read that
> "sti" is "protected mode virtual interrupt enable".
>
> I removed "sti" from ADLO because it is real mode. 
> Does It affect badly to my EPIA operation ?

sti should also work in real mode.
> ps.> If I reset my system with LinuxBIOS after booting it with PC BIOS,
> it works great.

Linux's reboot code sometimes doesn't work for some reason.
(I think it resets only CPU)

I wrote a small program to directly hit the power management device
found in EPIA to reset the system, not only CPU.
It works for me.
Changing 0x29 to 0x28 makes it power off instead of reboot.
See southbridge manual for detail. (0x4000 is PM base)
--
Takeshi

#include <sys/io.h>

main()
{
    iopl(3);
    outb(0xff, 0x4001);
    printf("01=%x\n", inb(0x4001));
    outb(0x20, 0x4005);
    printf("05=%x\n", inb(0x4005));
    outw(0xff, 0x4020);
    printf("20=%x\n", inw(0x4020));
    outb(0x29, 0x4005);
//    while ((inb(0x4001) & 1) == 0)
//        sleep(2);
}

[PATCH] Automatic RAM detection for VT8601

[ron minnich]

On Tue, 20 May 2003, SONE Takeshi wrote:
> Linux's reboot code sometimes doesn't work for some reason.
> (I think it resets only CPU)

the linux reboot code is very strange and I determined a few years ago 
that most of it doesn't work; rebooting seems to happen mostly through the 
triple fault scenario.
> I wrote a small program to directly hit the power management device
> found in EPIA to reset the system, not only CPU.

that's far and away the best way to reset a system.

The most acceptable way (nowadays) to integrate that into the kernel is as 
part of watchdog timer support. However it might be worth integrating into 
the power off/reset patches in freebios source tree.

ron