OKUJI Yoshinori <[EMAIL PROTECTED]> wrote:
> > Has this been changed since the old GRUB 0.5? It was there for a
> > reason.
>
> Yes, we have changed this to use BIOS, because your original code
> didn't work with the "floppy emulation" in the El Torito
> specification. It seems that we should reconsider the floppy probe
> technique. Perhaps the right thing to do is:
>
> 1) Call INT 13, AX=4B01 (Bootable CD-ROM - GET STATUS).
> 2) Check if the call succeeds.
> 3a) If succeeds, use the media type field of the Bootable CD-ROM
> Specification Packet to get the geometry.
> 3b) If fails, use your original probe routine to get the geometry.
That is good, but it isn't quite enough. LS-120 disks will still fail.
I propose:
1) Find BIOS geometry.
2) Check total size (or just number of cylinders).
3a) If under some reasonable size, say 6M or so, then use floppy
iterative probe.
3b) Else use normal BIOS geometry.
Just to be sure... what is the behavior of an LS-120 drive if booted with
an LS-120 disk in it? And if it can be changed, does the BIOS reported
geometry change or do we have to make a smarter probe? (which I would
have an idea for, but why worry about it if it doesn't happen?)
If an LS-120 drive would report the size of an LS-120 disk when booted
that way even when a floppy disk is inserted afterward, then the probe
shoudl be changed to try some sizes, then check to see if it doesn't
fit the sizes at all, and in any case, checking the number of available
cylinders in some way.
Could someone with an LS-120 drive report on what it does in the
circumstance mentioned just above?
--
Erich Stefan Boleyn \_ <[EMAIL PROTECTED]>
Mad but Happy Scientist \__ http://www.uruk.org/
Motto: "I'll live forever or die trying" ---------------------------
