Hacksaw (hacksaw@user1.channel1.com) writes:
> 
> Past experience seems to indicate that Phoenix BIOS does the wrong
> thing when reporting the boot drive unit to the boot code.

Thank you!

Yes, the HP Vectra XM uses a Phoenix BIOS.  When I recompiled the
boot code with -DDEBUG, I got "1 2 3 Read error" which demonstrated
that the fail occurred when trying to do the 1st read.  So I forced
the drive number to be 0x80 (wd0) and now it boots like a charm.

BTW: in start.S, does anyone know why I have to insert "addr32" or
"data32" (doesn't seem to matter which I use) after:

    movb    $0x80, %dl      # force hd0; PhoenixBIOS bug (bmw)

to make the compiler happy?  I last did 80x86 code pre-Pentium, and
I don't know what's up with these prefix(?) bytes.


In practical terms, since there are a helluva lot of machines using
the Phoenix BIOS, does this not represent a bug of sorts?  I think
it could be easily fixed by having separate binaries for fdboot
and wdboot instead of linking them all together.

Thanks all!

-bmw