When booting with the Red Hat 6.0 floppy, the first and second IDE controller (ide0 and ide1) are properly detected and both my IDE disks (/dev/hda with Win98, /dev/hdc with Linux), too, as well as the slave Atapi CDROM on the second IDE (/dev/hdd). However, it also erroneously "detects" a "non-IDE" device on /dev/hdb (slave on first IDE): I have no device physically or logically (BIOS) attached on the slave port of the first IDE. The errors are (Alt-F2): "<4>hdb: non-IDE drive, CHS=621/128/63" ... "<3>hdb: INVALID GEOMETRY: 128 PHYSICAL HEADS?" And when in disk druid/fdisk: "An error occured reading the partition table for the block device /tmp/hdb. The error was: Device not configured." It does not seem to influence the installation process and operations after installation, although the "<3>" error keeps popping up after installation and after baking a new kernel.
What brand of motherboard is this that is having these problems. Also, verify that you have the slace device on the ide0 chain set to "none" in the BIOS.
Motherboard: Intel "Advanced/ZP Baby-AT Board" (says on the cover) with a 82371FB PCI ISA/IDE Accelerator (PIIX) and 83437FX Pentium PCI bridge (Triton). Ordinary Pentium @ 75 MHz, 32 Mb edo ram. Yes, when I enter the setup of the BIOS: Primary IDE Master: ST5180A Primary IDE Slave : Not Installed Secondary IDE Master: QUANTUM FIREBALL_T Secondary IDE Slave : GCD-R540C And when ordinarily booting, only the these three devices are listed on the screen.
We do not have that particular board in the lab and also have not noticed this type of behavior with other boards. Just curious, does this still occur if you completely clear the CMOS and choose BIOS defaults without performance enhancements?
Thanks for your answer, but i'm sorry, i'm somewhat reluctant to clear the CMOS etc. As said, the "error" seems harmless; and i still don't understand why during BIOS bootup the "ghost drive" does not show up, and Windows 98 also does not mention or complain about it. I think it has to do with, and i agree it is not common, the fact that the slave is on the second ide ctrl and the first does not have a slave. I reported this error(?) cuz i thought it might signify something. I wouldn't mind if this bug is closed.
If you can try typing at the LILO boot: prompt linux hdb=noprobe and see if kernel still reports the drive being detected. If it helps then add to your /etc/lilo.conf file in the linux section append="hdb=noprobe" and rerun /sbin/lilo before rebooting.