AW: AW: [syslinux] flashing bios & memdisk

[H. Peter Anvin]

Johannes Feigl wrote:
>>No.  Big real mode is not real mode, and in particular 
>>MEMDISK doesn't have the control over what mode transitions 
>>the world outside MEMDISK performs.
> 
> 
> i understand.
> if only dos will be used (without himem, emm386 or something like that)
> that the cpu will be leaved in real mode, is that correct?
> and e.g. bios update would work...
> 

You don't know that.

It's possible that what breaks it is that the particular flash program 
does use big real mode, and doesn't consider that something else may 
break that -- in effect the author of that program did the mistake that 
you're proposing.

The big problem is that there isn't an easy way to test if you're 
already in big real mode or not, unless you have access to SMI, which is 
system-dependent.  Although you can trigger a #GP, because of some 
unbelievable stupidity on the part of the people designing the original 
IBM PC, #GP can't be easily distinguished from an IRQ 5 hardware 
interrupt.  The consequences of mistaking would be disastrous.

It would be possible -- in fact, fairly easy -- to hack the "raw" mode 
of MEMDISK to drop the system into big real mode after finishing instead 
of real mode; such a "bigraw" mode might solve the problem with this 
particular program.

The PC hardware... gotta love it.

	-hpa