Folk,If this isn't the best place to ask, please tell me (politely!) where to go instead.
I have Dell GX270 with a P4 Core2 and the Intel 82801EB chipset, which has been running Debian Linux (recently 2.6.16.11) happily for several years. The primary IDE drive (hda) is dying of bad blocks so I bought a WD 640GB SATA drive.
The SATA BIOS in this m/b always assigns the SATA drives to drive codes 0x80, etc, and the IDE drives after that, which caused some headaches with Lilo until I found the right options. I have a root f/s on the SATA drive (/dev/sda1) with / boot set up, and Lilo
is loading the kernels (2.6.16.11 and 2.6.26.5) just fine.When either of these kernels uses /dev/hda1 as root, they work fine, and mount the
SATA drives listed in fstab on bootup.When the same kernel is loaded using /dev/sda1 as root, it finishes initializing devices and frees the kernel memory used during bootup, then says "unable to create initial
console" on the first access to the SATA disk.Looking at dmesg from a successful bootup, the next thing after freeing memory is to load some more modules, so I de-modulised everything that should be required on bootup; only non-core things like SMB etc are still modules. So I don't believe I have
any unloaded modules required for SATA support.The 2.6.26.5 kernel has new support for the Intel ICH5 chips (Serial ATA(prod)... ->
ATA SFF -> Intel ESB, ICH, PIIX4 ...) enabled.The older kernel is using the SCSI_SATA and SCSI_ATA_PIIX config settings and has
BLK_DEV_PIIX enabled as well.The output of "lilo -T geom", my lilo.conf, output of "lspci", the two config files, and the output of dmesg from a successful (hda1) bootup are attached. Please let me know
what else I should provide. Clifford Heath.
Attachment:
lilo-T-geom
Description: Binary data
Attachment:
lilo.conf
Description: Binary data
Attachment:
linux-2.6.16.11.config
Description: Binary data
Attachment:
linux-2.6.26.5.config
Description: Binary data
Attachment:
lspci
Description: Binary data
Attachment:
dmesg
Description: Binary data
