In article <[EMAIL PROTECTED]> you write:
>From: Brian Brunswick <[EMAIL PROTECTED]>
>Subject: Grub error messages
>Date: Thu, 22 Apr 1999 11:29:54 +0100 (BST)
>
>> gcc version 2.7.2.3
>> GNU assembler version 2.9.1 (i686-pc-linux-gnu), using BFD version 2.9.1.0.4
>>
>> This builds a totally wrong stage1, with no warning at all. Code
>> offsets wrong etc relative to downloaded binary versions. I would
>> suggest putting in something to prevent this trap!
At least a warning in the README that early binutils will fail silently.
I had actually read the bit saying one needed 16 bit support, but
assumed initially that if it built cleanly, it was ok.
If this binutils went into redhat 5.1, it must be quite common.
>
> I don't think so. You should know 2.9.1.0.x is BETA release. Why
>must we check it even if it is a beta release? If you send a patch for
What! Oh well. I'm getting less and less happy with redhat and more
and more persuaded towards debian, which at least seems to have some
sort of standard of quality. Too many redhat things seem to be broken
out of the box, and never tested.
>it to us, we will be pleased to merge the patch, otherwise we will not
>work on it because it's not important at all.
>
I'd have to learn how to write configure patches - but I suppose that
will come in useful. I might give it a try if time permits, perhaps
find a single instruction that goes wrong I suppose.
>> Since then, in a very few minutes at odd times over the last couple of
>> weeks I've still been fighting with grub trying to get it to work. I
>> can install enough on my hard disk to boot to the grub command line,
>> but I can't get it to read the menu.lst file.
>
> Please write down the command that you used for the installation
>exactly.
Ok, (from memory!) its
install=(hd0,1)/boot/grub/stage1 d (hd0,1) (hd0,1)/boot/grub/stage2 0x8000 p
or
install=(hd0,1)/boot/grub/stage1 d (hd0,1) (hd0)1+16 0x2000 p
Having previously done (from linux)
dd < /mnt/boot/boot/grub/e2fsstage1_5 1<>/dev/hda seek=1
/mnt/boot is /dev/hda2, a normal linux ext2 partition.
The file completion works within it, so presumeably grub is happy
reading it.
The stage2 version gets to the command line ok, but fails to read
menu.lst. I've checked that it has had a divice word of 01ffff00 poked
into it.
I've even tried manually setting the values in stage 1 and entirely
manually installing it. (Someone should update the docs that the
fixed offsets have changed, BTW.)
The stage1_5 version even fails to get that far.
>
>> One thing thats really hindering is its silence. Surely it would be worth
>> having each stage at least announce itself as running, so one can see
>> whats going go. At least a Grub 1 1.5 2.. type of output. And some
>> kind of message about the menu file.
>
> Agreed. But I think the worst point should be that GRUB-installation
>is not user-friendly. So I have been considering changing the
>command-line interface heavily, though that will lead to
>incompatibility.
Agreed, but some messages are still worth while.
