Baslinux mailing list

[sindi keesan <keesan AT sdf.lonestar.org>]

Compaq Deskpro progress (it works, mostly)

> If I add emm386.exe to config.sys I can boot BL3 loop from the IDE drive
> using my scsi kernel and initrd.fs, and then mount the partitions on the
> scsi drive (and presumably transfer linux to them that way).
>
> If I try to boot with my scsi replacement for initfs.gz, from either the
> scsi drive (without IDE drive in there since scsi drive DOS is D:) or the
> IDE drive, I get the same failed boot:
VFS:  Mount root (ext2 filesystem)       [this is /dev/loop0]

> EXT_2-fs:  unable to read superblock   [can't read scsi superblock?]
> FAT:  unable to read boot sector

> On two other computers which did not come with scsi controller, my
> scsifs.gz replacement for initfs.gz works okay (to boot from IDE drive
> with the scsi kernel).

I was able to temporarily stick in an IDE drive as primary master, boot 
BL3 loop from it with regular initfs.gz, and copy BL3 and BL2 to ext2 
partitions on the scsi drive, and they boot fine with loadlin.

Xvesa in ext2 appears to work (I can get into icewm) but on exit the 
screen is scrambled (blue on the bottom half) from icewm, or 20 columns 
with colors (from plain Xvesa) and the computer reboots itself.  When I 
boot loop linux Xvesa simply refuses to load, as on the Presario.  We have 
only one ATI video card which works right with Xvesa and loadlin, th 
newest one.

I think this computer simply is not going to work with loop linux that 
tries to mount /dev/sda1.  If I want to do rescue I will have to put in an 
IDE drive with loop BL3 on it, or redo BL3 2-floppy version somehow to use 
my scsi kernel - how would I do this?  I don't need X or a browser for the 
rescue disk, probably just busybox.  A newer e2fsck and related library 
might be helpful to rescue BL2.  Maybe I can put together a modified 
2-floppy BL2 set with these from the full BL2.  BL2 ramdisk needed a newer 
ext2 library to work with e2fsck from BL2 ext2.

Someone else got similar error messages on hardware that worked with RH 
7.2 but not 7.3.  Some new bug?  Maybe I should compile an older scsi 
kernel some day.

Sindi