[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: GRUB



Hi.

In article <-7HZb.A.3TC.q3u13@murphy>
 gearhead@twcny.rr.com (Steve Przepiora) writes:

> Disk /dev/hda: 255 heads, 63 sectors, 1401 cylinders 
> 
> I dont have a problem with disks greater than 1024 cylinders.
> BTW its some older BIOS's that have this problem, not lilo.

LILO does have problem with very huge disks whose number of
sectors does not fit in 24bit addressing, at least in ver 21.

If you use over 8GB diks (ex. 10GB or more) and place the linux
kernel over the 24bit's wall, then lilo can't boot up.

Please check the first.S and second.S in lilo's source code.
lilo uses BIOS via int13 with AH=02H which is traditional
CHS access method, even if you set the "linear" option for lilo.

When you switch the BIOS to use "LBA" geometry translation, 
you can escape the 1023 cyl's wall at 504MB, but can not for 
24bit's (CHS addressing) wall.

There comes the "LBA addressing" (int13 AH=42H for read)
recently with some advanced BIOS. And the most recent development
version of grub (I have 0.5.92) uses this function. following is
the sample from the code:

        /* check if LBA is supported */
        movb    $0x41, %ah
        movw    $0x55aa, %bx
        int     $0x13

        jc      lba_probe_error
        cmpw    $0xaa55, %bx
        jne     lba_probe_error

        /* get the geometry (limited to 2TB!) */
        movb    $0x48, %ah
        movw    $ABS(drive_parameter), %si
        movw    $0x1a, (%si)
        int     $0x13

From the ChangeLog of it, it seems that the stage1/stage1_lba.S
added at 1999-07-11 by OKUJI Yoshinori, a Japanese GNU hacker
(he is a translator of "Brave GNU world" also).

This stage1_lba.S is now under development actively, but OKUJI
(btw, OKUJI is his family name, and Yoshinori is his personal name)
told me that "as a developer, I am grateful if you test the code. 
but if I was a mere user, I do not recommend you to use this.
LBA is totally under development, and there are many broken BIOSes 
on this standpoint. Let's try if you prefer an adventure."

So I do not think we Debian just move grub NOW for coming stable
release, though I hope we can use it soon. and I am happy if we Debian
provide the good & handy documents for it and the nice setting tools.

btw, anyone tried my "extipl" package for potato ? I think it is
superior than current i386 "mbr" on features that it can boot up
the system on the other hard disk (if the OS in that place can do, i.e. 
the MS-xxx can not work. and the default is limited to the first disk)
can change default boot partition on the first disk at booting time, 
and can show the some information about the partitions on the disks at
booting time.

I thought Debian can use this extipl in place of "mbr" for i386, 
but I feel that this i386 only topic will not catch the attention
from people on current debian-boot discussion, so I may have not 
told about this on debian-boot ML,,,

-- 
  Taketoshi Sano: <sano@debian.org>,<sano@debian.or.jp>,<kgh12351@nifty.ne.jp>


Reply to: