DOS ain't dead

> Hi,
> I just upgraded to linux kernel 3.4.4 and tested loadlin 1.6e that should
> have fix for booting newer kernels but it seems that doesn't work.
Hmmm. I guess you know kernels after 2.6.30 added support for LZMA and BZIP2 in addition to zlib.

Clearly your kernel decompressed successfully and worked with linld. So I have wondered what might have been the problem. Having looked at linld for the first time I noted it used a stripped modified version of the loadlin source and was written due to issues with loadlin! :) Did you try booting with loadlin's -v,-t,-d logging options? If not see doc/manual.txt in http://youpibouh.thefreecat.org/loadlin/ loadlin-1.6e.tgz

> Kernel and initrd (about 3+3MB) decompressed and followed by immediate restart
I wondered if what you ran into was a mode switching issue during a Loadlin callback? Did you try with the undocumented -f switch? See below:
5.1 Problems detecting V86
----------------------
On some 486 clones we have problems with CR0 while probing for V86 mode.
(as reported by Jacek Zapala [snip] ).
It can happen that the CPU is in realmode, but PAGING is enabled !
This is possible, but neither documented nor supported by INTEL.
May be the motherboard's BIOS is mapping shadow ram this way,
or one of those old and strange EMM managers is used,
but probably this is an indication of a not exactly compatible
486 clone. The -clone switch byepasses the CR0 check and assumes
V86 if an EMM manger is found.
But of course this EMM manager must not use real paging !

On the other hand, if LOADLIN detects V86 _and_ you are sure that the
machine is in _real_ realmode, then you may use the -f switch
in order to force LOADLIN to 'detect' realmode.
NOTE: This option intentionaly is not mentioned elsewere, because
it realy is dangerous.

Might be worth a shot if anything to just understand where loadlin failed???


> Fortunately linld 0.97 works fine so forgot loadlin crap...
Until your post I wasn't familiar with linld (well don't remember it), so thanks you for the info. I really like its small overhead and have also now used it to partly boot a test kernel which loadlin bombed out with. Admittedly my use of both was under DOSBox... So I wasn't surprised either :) I am familiar with Dosbox's boot, imgmount - but have had my reasons....;)