Re2: LOADLIN and 2.4 kernels

Re2: LOADLIN and 2.4 kernels

[Floydsmith]

>> >Hi everyone,
>> >
>> >I got a bug report of LOADLIN not working with recent -ac kernels, and
>> >thought it might have something to do with my recent A20 changes that
>> >were added to -ac.  However, in trying to reproduce this bug, I have
>> >been completely unable to boot *any* 2.4 kernel with LOADLIN-1.6,
>> > trying this from Win98 DOS mode.
>> >
>> >Anyone have any insight into this?  I really don't understand how the
>> >A20 changes could affect LOADLIN, and it's starting to look to me that
>> >there is some other problem going on...
>> >
>> >        -hpa
>>
>
><snip>
>
>> loads the 2.4.x kernel into a buffer. The kernel then attempts boot
>> just the "boot" sector stuff. This again probes for the total amount of
>> system ram (64MB). But, because of the much greater size of 2.4.x
>> kernels some memory location that himem uses (I think - maybe BIOS

>Sounds like something booting to Safe Mode Command Prompt Only would fix, 
>as opposed to booting to plain command prompt mode
>command prompt mode will load some drivers (such as himem), better not to 
>load them when using LOADLIN. Safe Mode Command Prompt Only boots 
>straight to the command prompt, very similar to setting init to /bin/sh 
>for a completely bare single-user mode.


Yes, indeed, not loading himem does solve the problem I had. But, do to the 
fact that I need extented memory (for a DOS ramdisk) and for some TSR(s) 
(like smartdrv) for a LS-120 boot disk I use as both a Linux and DOS "rescue" 
disk, I need "himem".

Floyd,

Re: Re2: LOADLIN and 2.4 kernels

[H. Peter Anvin]

Followup to:  <51.10cc64a8.28cb50f8@aol.com>
By author:    Floydsmith@aol.com
In newsgroup: linux.dev.kernel
> 
> Yes, indeed, not loading himem does solve the problem I had. But, do to the 
> fact that I need extented memory (for a DOS ramdisk) and for some TSR(s) 
> (like smartdrv) for a LS-120 boot disk I use as both a Linux and DOS "rescue" 
> disk, I need "himem".
> 

Someone reported to me that LOADLIN consistently fails to load kernels
> 1 MB both with and without A20 changes; probably it tries to use the
old "sector count" field in the kernel.  This would explain what we
have observed, and so I consider this matter closed as far as I'm
concerned.  LOADLIN needs to be updated to continue to be useful, and
I don't know if Hans is still around, even.

	-hpa

-- 
<hpa@transmeta.com> at work, <hpa@zytor.com> in private!
"Unix gives you enough rope to shoot yourself in the foot."
http://www.zytor.com/~hpa/puzzle.txt	<amsp@zytor.com>