Strangely enough, the bios is definetly positive about his controller
being a MFM/RLL one. Anyway, this would explain why the disk runs (if it
is in fact an IDE disk on an IDE controller) and why linux boots; but
I then do not understand how the bios recognizes the controller, which
it indeed does (lilo works, msdos used to work on this machine).
Also, the C/H/S trickery might be solved by the translations made by the
drive itself.
Oh well ... it works, linux is great, ip packets are routed ... what more
would one want ?
--Thomas Pornin