Hello OKUJI Yoshinori!

 On Sat, 7 August 1999 at 12:21:16, you wrote:
 >   Dirk, send any mail about GNU GRUB to [EMAIL PROTECTED] instead of
 > debian-hurd. I will not tell anyone this any more, and I will not
 > reply any mail if you do ignore my advise.

Well - Grub still is rather Hurdish so I just carelessly postet this
to the list in order to let others know about possible problems.
I did not intend to make this a bug report, but you will get what
you ask for...

(Please note that I used grub from the hurd 0.0 release up to date
 - it replaced LILO and I never, ever did regret it, even though
 my mother also happens to be called Lilo (the short form of her
 real name Lieselotte). You see - I am a very happy Grub user.)


 >   I'm not sure, but SCSI drives normally support LBA mode. So I
 > suspect that you just did something wrong. Please tell us what you did
 > in details.

My Hardware:
------------
- ASUS P2-DS configured for 1.4 MPS spec, 2 CPU's found by GRUB
  with 'impsprobe' (the board is based on Intel P2X4E and Intel
  440BX Chipsets)

- 393208KB /proc/kcore, all detected by GRUB

- on board Adaptec AIC-7890 U2W LVD SCSI Controller, configured
  to provide translation for drives > 1GB (I probably do not need
  this because I don't do proprietory stuff anymore.)

  Attached devices:
  Host: scsi0 Channel: 00 Id: 00 Lun: 00
    Vendor: IBM      Model: DDRS-39130D      Rev: DC1B
    Type:   Direct-Access                    ANSI SCSI revision: 02
  - this is an LVD device
  - cfdisk 0.8 reports C/H/S = 1111/255/63 under GNU/Linux 2.2.0

  Host: scsi0 Channel: 00 Id: 01 Lun: 00
    Vendor: PLEXTOR  Model: CD-ROM PX-40TS   Rev: 1.00
    Type:   CD-ROM                           ANSI SCSI revision: 02

- on board IDE ports (U-DMA/33 capable):
  Attached devices:
  - 2x IBM-DHEA-38451, both configured as masters on port 0 and 1
    respectively, C/H/S = 16383/16/63, 472k cache

- On my system(s) (hd0,0) traditionally is a small partition dedicated
  to GRUB and some kernels to choose from; recently I started to
  mount this under "/boot", making it a rather clean and reliable
  solution for my purposes.

After I compiled from the 'alpha' sources I installed stage1_lba
in the following ways:

(hd0) configured as SCSI disk, (hd0,0) with ext2-fs:
install= (hd0,0)/grub/stage1_lba d (hd0) (hd0,0)/grub/stage2
         0x8000 p (hd0,0)/grub/menu.lst
- does result in "Error: LBA", requires Ctrl-Alt-Del

(fd0), ext2-fs:
install= (fd0)/grub/stage1_lba d (fd0) (fd0)/grub/stage2
         0x8000 p (fd0)/grub/menu.lst
- does result in "Error: LBA", requires Ctrl-Alt-Del

(hd0) configured as IDE disk, (hd0,0) with ext2-fs:
install= (hd0,0)/grub/stage1_lba d (hd0) (hd0,0)/grub/stage2
         0x8000 p (hd0,0)/grub/menu.lst
- works, i.e. stage2 is loaded (this shuts a lot of guesswork
  out I suppose, so I gave it a try today)

The stage1 worked for both situations where stage1_lba failed.
 

 >   No. In fact, I have been using the emulator to install GRUB into a
 > floppy disk when developing GRUB.

You are right - it does work.  Stupid me - it has been late
and I guess I probably tried to acess a floppy disk that was
lying somewhere on my cluttered desk - inserting it into the
drive definitely helped to overcome the nasty problem. ;-)
Did I hear someone laugh?

Yet another issue - is scrolling of lists with more than 12
menu entries supposed to work?  I tried this with 0.5.91 and
it did not scroll, but as far as I can remember one still
could select items that exceeded the field if one remembered
the keystrokes.  The intention was to have a floppy with all
sorts of weired combinations of install commands handy.
Of course - one can type them in or one could edit an entry
matching closely the situation, however, this is rather
inconvenient if one has a german keyboard layout.


Well - now for the fun part:
----------------------------
You should modify the examples for various graphic adventure games
that are quite frequently found on many PC-based playstations:

- Loading OS/2 (v3.0, also known as Warp) from an HPFS formatted
  partition worked nicely, just use "root=(hd?,?)" and then
  "chainloader= +1"  (No - I am unable to ever test this again
  since I nuked all proprietory stuff.  Perhaps one needs to use
  "rootnoverify" instead, but even "root" should do.)
  The OS/2 example definitely needs to be changed since I have
  never seen a chainloader that would be in any way specific
  to this system.  Developing one would be a waste of time since
  the blocklist notation works, provided it worked earlier together
  with the braindead bootmanager that comes with OS/2 of course,
  but I nuked that bootmanager thingy after I once discovered that
  Grub can do the job as well.  Prior to that date I chainloaded
  Grub from the OS/2 bootloader.  BTW - it also works the other way,
  i.e. one can chainload OS/2's bootloader from Grub as well of
  course. Finally - if anyone really wants to get it working under
  adverse conditions one could tell people to install OS/2 on an
  ext2-fs -- someone made it work with it and he even provided some
  means to boot it cleanly, i.e. he did not rely on some mysterious
  piece of bootloader code at the beginning of a partition. He made
  it all available under the GPL so it would be politically correct
  to mention this. (OS/2 has the feature of installable filesystems
  - someone just took the ext2 sources and made them work under OS/2.
  I tried it.  Since then I know for sure that icons are not the way
  to go if one has to manage deeply nested directory structures
  together with lots of files and/or subdirectories.)
- Using 'makeactive' never seemed to be needed, so I neither used
  it when I had OS/2 installed on my own box, nor on a box of a
  friend (primary partitions only, I never used anything else)
- Using 'makeactive' seems not to be needed with various versions
  of DOS as well - I tried "chainloader= +1" with Novell DOS 7.0,
  DOS'95 and DOS-NT -- there have not been any problems so far
  (primary partitions only, I never used anything else)

I could offer to work on the GRUB documentation but you would have
to show some patience since I have a rather demanding full time mac
job(1).  I could also offer to create and maintain a german translation.

(1) I happen to produce cylinder heads (V6/V8 engines for Daimler
    Chrysler AG) - perhaps this explains some random stupidity
    on my side when the topic is computers.

/bye
Dirk

P.S.: My favorite - "color= 0x78 0x4f".  Cool! :-)

Reply via email to