On Sat, 6 Nov 1999, OKUJI Yoshinori wrote:
> OK, I'll add this, as it seems to be necessary to initialize APM in
> the real mode. The problem is that I can't get the specification from
> the net. I sent a mail to Intel, but I don't expect that they will
> answer my message. If you know where to find it, let me know.
Have a look in linux arch/i386/boot/setup.S, in the #ifdef CONFIG_APM
section; various real-mode calls are used to work out whether an APM
BIOS is present (int 0x15 AH=0x53) and whether it supports 32-bit
mode. We store away a block of parameters obtained from the BIOS (BIOS
code segment, entry point offset, 16 bit code segment, data segment,
code segment length, data segment length, APM BIOS version, APM BIOS
flags) - this is all that really needs to be present in the Multiboot
information.
Of course, if we check for an APM BIOS and find there isn't one, we
should still set the flag bit indicating the fields in the multiboot
info structure are valid - it's better to know that there is no APM
BIOS than not to know whether one has been checked for.
> > MCA bus feature table (also not very useful now)
>
> Is it enough to copy the table returned by INT 15 AH=C0h?
Yes, I believe so.
> > Flag to indicate whether a PS/2 mouse is present
>
> I don't think this is necessary. Look at the file pc_keyb.c in Linux
> 2.2.x.
You are right.
> > It might also be useful to add a string describing the bootloader, so
> > that the operating system can log it.
>
> Can you give me an example?
Just something like 'GRUB 0.5.13' or 'mknbi-multiboot 0.12' or
something like that; a string that the operating system can log so
that we can find out how it was booted.
Steve Early
