[syslinux] Initial bootloader requirements of extlinux

Shao Miller Shao.Miller at yrdsb.edu.on.ca
Sat Nov 14 14:26:22 PST 2009


Simon Rowe wrote:
>
> that again shows SI being set up. I had hoped to come across something 
> more
> official looking, like the T13 doc about the equivalent GPT behaviour.
>

They might enjoy: http://wiki.osdev.org/MBR_(x86)

If I have my facts straight (I might not), this is De Facto standard 
behaviour since DOS has relied on it historically.  So I'm not sure that 
it's in any industry standards document.

You might notice, however, that in industry specifications for INT 0x13 
(such as EDD), some of the fields in an "IBM-style" partition record 
line up with what certain low INT 0x13 functions (such as AH=0x0A) are 
expecting in registers upon the call.  This would suggest that the 
purpose for DS:SI pointing to such a record is so that some future code 
already "knows" where to start reading sectors from for their INT 0x13 
calls.  This seems quite natural to me.

For example, compare the expected register entry states in section 8.7 
of the EDD-4 spec. draft with 
http://en.wikipedia.org/wiki/Master_boot_record#MBRs_and_disk_partitioning  
In fact, the convention for having the active partition marked with a 
0x80 byte corresponds to the expected DL contents for an HDD boot of the 
first hard disk drive.

Now is there some way that a series of bytes like this (a partition 
record) could automatically populate the contents of the registers 
expected by INT 0x13, AH=0x0A?  If I recall correctly, POPping takes 
successive values from memory...  DX, CX...

- Shao Miller



More information about the Syslinux mailing list