I was able to get FILO up as a PAYLOAD for Linuxbios in a very short time. And it was able to load and start executing either memtest86 or the Linux kernel off of an IDE disk using GRUB.
Unfortunately when Linux is booting up and attempts to mount the root file system, it chokes complaining that it can't mount the root device (the entry is fine, and linux reports the root device properly as it tries to mount it).
When I stick back in the stock Phoenix BIOS, the same disk/kernel boots up just fine and has no problems whatsoever mounting the root device.
I suspect that maybe FILO has done something naughty to the IDE registers, as I think the root mount is the first work the kernel is doing on the disk. The kernel can see the CDROM (as well as all the other non-IDE devices).
Anybody have any ideas on this?
Craig C. Forney Opus Innovations LLC
On Fri, Oct 10, 2003 at 02:13:00AM -0700, Craig C. Forney wrote:
I suspect that maybe FILO has done something naughty to the IDE registers,
I think the IDE driver that FILO uses does only very basic operation to the IDE.
However LinuxBIOS might set up the IDE controller as different way than Phoenix.
Could you use Etherboot as payload, and do a network boot (not IDE boot, it uses almost same driver) to see what happen?
as I think the root mount is the first work the kernel is doing on the disk. The kernel can see the CDROM (as well as all the other non-IDE devices).
Partion table is the first thing the kernel reads from disk. Does it go well?
Anybody have any ideas on this?
Also, if you were using /vmlinuz, try ELF boot instead (convert your kernel with mkelfImage from LNXI). bzImage loader could have bugs.
-- Takeshi
* Craig C. Forney cforney@opus.com [031010 11:13]:
Unfortunately when Linux is booting up and attempts to mount the root file system, it chokes complaining that it can't mount the root device (the entry is fine, and linux reports the root device properly as it tries to mount it).
Are you using an initial ramdisk to load the rootfs driver or fs when booting with grub?
Stefan
On Fri, 10 Oct 2003, Craig C. Forney wrote:
Unfortunately when Linux is booting up and attempts to mount the root file system, it chokes complaining that it can't mount the root device (the entry is fine, and linux reports the root device properly as it tries to mount it).
can you send the messages. We used to have this type of problem, and it was almost always partitioning issues or block numbering issues. Linux acts a little differently when there is no BIOS information describing the drive.
I suspect that maybe FILO has done something naughty to the IDE registers, as I think the root mount is the first work the kernel is doing on the disk.
dubious. I'm pretty sure that's not it.
ron