> Also, there seems to be a problem booting from internal LS-120 drives. > Normal operating systems (Windows...) can boot from internal drives if > the BIOS can handle it. GRUB hangs when I try to boot from an ordinary > 3,5" disk inserted into the LS-120. I've tested GRUB 0.4, but not > further explored this issue. It may be a problem of GRUB, but perhaps it > is a problem of the chipset or the BIOS. I will give a detailed report > once I've tested it with the newest version of GRUB.
I wrote about this problem earlier in debian-hurd (2 Dec). I hacked my copy of GRUB so that it works from my internal LS-120. I'm copying the relevant paragraphs again, below. By the way, SYSLINUX boots off my LS-120 without problems: it applies a different algorithm for deciding whether the device is a floppy or a hard disc. Boot sectors can be quite amusing, really ... Edmund As I wrote before, the GRUB boot floppy provided with gnu-0.2.tar.gz doesn't work off my LS120 IDE floppy. Neither does the 0.5 release of GRUB, which I built from source. However, after many failed experiments, I managed to modify grub-0.5's boot sector so that it works with both my ordinary and my IDE floppies. (I include the patch below, and I will tell Erich about my experiences.) ... My GRUB patch: It's just a hack, not a serious attempt to improve the software, but it did give a floppy that boots off both my ordinary and my IDE floppy drives. In grub-0.5/stage1/stage1.S the patch is: < andb $BIOS_HD_FLAG, %al < < /* jz floppy_probe */ < .byte 0x0f, 0x84; REL(floppy_probe) --- > orb $BIOS_HD_FLAG, %al > > /* jz floppy_probe */ > .byte 0x0f, 0x84; REL(floppy_probe) If you prefer to patch the binary boot sector, the change is: < 24 80 --- > 0c 80 That's at address 0x55 in grub-0.5/bin/stage1; the address is different, no doubt, in the Hurd's boot floppy, which is based on an earlier release of GRUB, but the same change might work for that too. Edmund
