Re: New MBR code, LBA capable

[Thorsten Glaser <tg-2003m@netcologne.de>]

(A summary is at the end of the mail)

Me, earlier this day:

>I've just coded an LBR capable Master Boot Record replacement
>because I was bored.

I've now finished working for today.
Results are attached:

 - mbr-101
	The new MBR code, second version
	Fixes only optical problems
 - fdisk-101
	Sync with mbr-101
 - biosinsboot
	Diffs for biosboot and installboot
	see below
 - boot-diff
	Diff for /boot, see below

Status:

MBR:	works fully, boots all kinds of operating
	systems I can find on both CHS and LBA (I
	even created a test partition for LBA)

biosboot: can boot /boot from CHS like before
	can be loaded by MBR from CHS like before
	can be loaded by MBR from LBA
	can load /boot from pure LBA into memory

installboot:
	is capable of writing the "new" biosboot's
		block table (5 bytes per entry instead
		of 4) for both CHS and LBA

boot:	This is the only thing I could not get
	working. I have attached preliminary
	diffs as well, but the #if 0'd code I
	marked does not compile when switched to
	#if 1, and I have no experience with
	in-C-file-assembly and gcc.
	Maybe Toby can shed some light on that.

----------------------------------------

Summary:
 - we can now use "our" MBR as boot loader and
   boot manager for CHS and LBA
 - biosboot/installboot are LBA enabled now, but
   this leaves fewer entries for /boot blocks.
   They can be used to load a /boot from LBA now
 - /boot can still only load kernels that reside
   in CHS zone, and the disklabel has to be there
   as well.

//Thorsten
-- 
Willst Du wegen dummer User immer 'Ja, ich will' nach einem rm an /dev/tty
eingeben müssen?		-- Bodo Eggert in de.alt.sysadmin.recovery